We made the cover!
This shows immune cell subtypes interacting w cancer cells in breast cancer metastasis, via spatial proteomics (IMC) of a biopsy. Overlaid with signaling arcs. No genAI used
paper: doi.org/10.1158/0008-5472.CAN-26-1086
image: aacrjournals.org/cancerres/issue/86/22
@RoussosTorres @CR_AACR
Lior Pachter
Lior Samuel Pachter (Hebrew: ליאור שמואל פאכטר) is a computational biologist. He works at the California Institute of Technology, where he is the Bren Professor of Computational Biology. He has widely varied research interests including genomics, combinatorics, computational geometry, machine learning, scientific computing, and statistics. More on Wikipedia
-
Lior Pachter reposted
Adam MacLean
@adamlmaclean
-
Lior Pachter
@lpachter
Introducing Manimble: a free, open-source slide editor in which every slide is a Manim program (so underlying the deck is Python). It alows for editing like in Keynote or PowerPoint while seamlessly incorporating AI edits / animations. pachterlab.github.io/Manimble
1/🧵Post not available here, view on X
Lior Pachter
@lpachter
BioCartoon is integrated into Manimble so one can import directly from it, but it also exists as a standalone web or desktop app, or CLI. More on BioCartoon coming in another thread soone. 12/
Lior Pachter
@lpachter
13/ I've written and presented several talks with Manimble now, and have found it to be very liberating not to be constrained by Google Slides, PowerPoint or Keynote. Hope others find it useful and happy to incorporate any feedback (or fork and modify to your liking!) 13/13
Lior Pachter
@lpachter
Lior Pachter
@lpachter
Yesterday I posted about Manimble, a new open-source slide editor where every slide is Manim code. One of its cool features is built-in biocartoon.com: a free, open source library of 15,000+ biology icons, developed by @Josephmrich x.com/lpachter/status/2122000632643260887
1/🧵
Lior Pachter
@lpachter
Introducing Manimble: a free, open-source slide editor in which every slide is a Manim program (so underlying the deck is Python). It alows for editing like in Keynote or PowerPoint while seamlessly incorporating AI edits / animations. pachterlab.github.io/Manimble
1/🧵 -
Lior Pachter
@lpachter
Yesterday I posted about Manimble, a new open-source slide editor where every slide is Manim code. One of its cool features is built-in biocartoon.com: a free, open source library of 15,000+ biology icons, developed by @Josephmrich x.com/lpachter/status/2122000632643260887
1/🧵
Lior Pachter
@lpachter
Introducing Manimble: a free, open-source slide editor in which every slide is a Manim program (so underlying the deck is Python). It alows for editing like in Keynote or PowerPoint while seamlessly incorporating AI edits / animations. pachterlab.github.io/Manimble
1/🧵Post not available here, view on X
Lior Pachter
@lpachter
My own contribution to test this: the eight founder strains of the Collaborative Cross and Diversity
Outbred mice, scaled to one another by published body masses and released CC0. Search "founder mouse" in BioCartoon or Manimble and they're one click away.
10/
Lior Pachter
@lpachter
BioCartoon is free and open source. Amazing work by @Josephmrich building this. Try it and contribute your drawings! biocartoon.com
github.com/josephrich98/biocartoon
11/ -
Lior Pachter reposted
Michael Gandal
@mikejg84
Biorender 2.0 for free
www.biocartoon.com -
Lior Pachter reposted
Ryan O'Donnell
@BooleanAnalysis
I lectured on the "ETH false" (#3sat in subexponential time) result in my grad complexity class at CMU today.
Didn't have time in 80 mins to do the full proof, but I hope I got across the main plan, and some of the basic TCS ingredients going into it.
youtu.be/_HlQGMcQWhA -
Lior Pachter reposted
Dan Landau
@landau_lab
Faculty ad is out,
please help spread the word
Come #dreambig together, without a doubt,
in the best city in the world!
jobs.sciencecareers.org/job/681004/tenure-track-assistant…
Dan Landau
@landau_lab
🚨We're recruiting! Calling all bold & original thinkers! Folks with imagination, drive & the (rare) ability to find the thin seam where the fantastical and real meet. Come join a thriving science community, in, beyond a shadow of a doubt, the greatest city in the 🌎 @nyknicks
-
Lior Pachter reposted
The Dartmouth
@thedartmouth
BREAKING: Santiago Schnell has resigned as provost of the College, after a committee-led review of his academic publications and public writing found that he had used artificial intelligence without proper disclosure. www.thedartmouth.com/article/2026/10/weaverbell-klee-shar…
-
Lior Pachter reposted
Caltech
@Caltech
Thayer (Ted) Scudder, Caltech's only professor of anthropology for the latter half of the 20th century, passed away October 22 at the age of 96.
www.caltech.edu/about/news/caltech-mourns-the-passing-of-… -
Lior Pachter
@lpachter
This is quite something.
Ryan Williams @rrwilliams.bsky.social
@rrwilliams
An internal model at OpenAI has produced a claimed refutation of ETH! Russell Impagliazzo, Mohan Paturi, and I have been asked to help organize an open and independent examination of this claim. 1/4
-
Lior Pachter reposted
Thomas Yeo
@bttyeo
Potential misconduct in a high profile fMRI foundation model study: neuroskepticbsky.substack.com/p/the-sherlock-holmes-of-sc…
-
Lior Pachter
@lpachter
"This is a bounded comparison, not a claim to have classified all nine-ray coverings or established worldwide priority."
LOLSOB.
Tony Tanuki
@tonyyueyu
As math research is increasingly carried out by AI agents, effective communication and engaging exposition become ever more important for human participation in the process.
We introduce mathdb.com/papers at MathDB, where we welcome a wide variety of research and expository papers, which may be accompanied by videos, code, and Lean formalization. Community comments and reviews are also supported. -
Lior Pachter
@lpachter
I was just reading the paper "Genome diversity in Ukraine" academic.oup.com/gigascience/article/10/1/giaa159/6079618 which mentions Jews exactly once as part of a list of "smaller minority groups" that were "historically present".
There were 2.5 million Jews in Ukraine pre-WWII. 1.5 million murdered. Now ~30,000 -
Lior Pachter
@lpachter
Found out last night that Thayer Scudder has passed away. He was an Anthropology professor @Caltech, and I had the privilege of taking his class as an undergraduate. One of the best classes I've ever taken. He was an amazing man with great integrity. May his memory be a blessing.
-
Lior Pachter
@lpachter
The dire wolf Colossal fiasco was embarrassing. And, it turns out, a prelude to this...
TechCrunch
@TechCrunch
Warhound, the brain child of Ben Lamm and Palmer Luckey, emerges from stealth to genetically engineer dogs for military and law enforcement.
techcrunch.com/2026/10/22/ben-lamm-raises-200m-for-warhou… -
Lior Pachter
@lpachter
edgePython v0.5.0 released at github.com/pachterlab/edgePython (`pip install edgepython`)
Matches the forthcoming edgeR 5.0 with several extra features as well as some edgeR bugs avoided.
Major speedups (4-6x faster, camera >100x), I/O improvements, and proper support for kallisto. -
Lior Pachter
@lpachter
Nice to see someone porting kallisto to rust: github.com/deminden/kallistors
There is still a way to go in terms of validation and the gamut of kallisto features (+ bustools, lr-kallisto, kb-python etc.) but it's looking good. -
Lior Pachter reposted
Abhishek Saha
@ObhishekSaha
2/ About three years ago, when I barely used Twitter -- I think I had only 50 followers then -- I suddenly found my mentions blowing up.
He had decided, based on an article I'd written, that I was SEXIST; he had tweeted this out to his 20K followers; and tagged my employer. -
Lior Pachter
@lpachter
So after weeks of hearing about how mathematics is not about the results per se, but about the journey and the understanding that comes with it…
… the truth turns out to be that all along it was just about the results.
Alex Kontorovich
@AlexKontorovich
In the US, we graduate on the order of 1000 math PhDs a year, and there are about 300 postdoc jobs (and ~70 tenure track jobs three years later). So no, a math PhD is not, and has never been, a "gold medal" (in the sense of an academic position).
And is "merit" about potential or about actual results? I really hope it's not the former (which is much more error prone to measure)!
About Zhang specifically, there's much more to his story than what's been reported, and it's devastating how much he suffered before his breakthrough. -
Lior Pachter
@lpachter
This is a crazy take.
Yitang Zhang was "unknown" *because* math is *not* a meritocracy. He was kicked out of mathematics because his advisor didn't want his own errors uncovered. Other mathematicians didn't give a @$#!. Zhang lived in his car, worked at Subway.
Alex Kontorovich
@AlexKontorovich
This is the exact opposite interpretation of what actually happened. The case of Yitang Zhang perfectly illustrates how mathematics culture, at its best, is a pure meritocracy. Regardless of how “unknown” you may be, write a paper that proves a longstanding problem and we will read it, and if it’s right, then you will be instantly recognized as a hero by the community.
This part of mathematics may now be gone forever (due to AI), and this is very sad; this is one aspect of what so many of us are mourning, even as we look with awe at all the amazing discoveries we’ve already been gifted and many more sure to come.
Lior Pachter
@lpachter
Suggesting that somehow the acceptance of Zhang's Annals paper is evidence of "pure meritocracy" is absurd. It's chutzpah.
www.quantamagazine.org/yitang-zhang-and-the-mystery-of-nu… -
Lior Pachter
@lpachter
Grant Sanderson (@3blue1brown) is the most influential mathematician of the 21st century.
-
Lior Pachter
@lpachter
Very sad to learn today of the passing of James Lloyd. www.linkedin.com/pulse/vale-dr-james-peter-brook-lloyd-ja…
-
Lior Pachter
@lpachter
The @CollegeBoard: "AP Exams..enable secondary school students to demonstrate college-level achievement"
Also @CollegeBoard: "Do you want to figure out how quickly you’ll reach 10,000 followers on Instagram? ..AP Precalculus..can help you calculate your rise to influencer status" -
Lior Pachter
@lpachter
Rota lesson #4 (from the ten lessons he wished he'd been taught).
Vince Vatter
@VinceVatter
To the extent that mathematics is getting mogged by AI goodharting, it’s because mathematics goodharted itself into this position by rewarding the solving of conjectures over understanding, exposition, and pretty much everything else.
Lior Pachter
@lpachter
Lesson #5 also comes into focus now. The unreasonable effectiveness of AI is likely due, in part, to it not being hamstrung by just a few tricks.
Source: www.ams.org/notices/199701/comm-rota.pdf -
Lior Pachter
@lpachter
Why is it called the quasi-Riemann hypothesis if it’s a fact?
-
Lior Pachter reposted
🌊💙 Viking Resistance 💙🌊
@BlueCrewViking
Note that Terry Tao has now made substantial silent updates to the slides and seems to have listened to this particular criticism. Overall it’s still pretty out there with naive analysis, though.
-
Lior Pachter reposted
Rakesh Bhandari
@postdiscipline
I don't think anyone needs reminding of this Ben Shapiro video youtube.com/watch?v=X9FGRkqUdf8
-
Lior Pachter
@lpachter
The three-story house at 00:10 is for sale for $4.95 million. www.redfin.com/CA/Long-Beach/6519-E-Seaside-Walk-90803/un…
Jordan Hall
@JordanHallWX
Large swells slam into Long Beach, California as multiple homes hold on for dear life to not fall.
@MyRadarWX @MaxVelocityWX #cawxThis media couldn't be displayed. See it on the original post. -
Lior Pachter reposted
Joshua B. Plotkin
@jplotkin
Opportunity: Penn MindCORE Fellowship!
MindCORE seeks to recruit outstanding postdoctoral researchers for our Research Fellowship for Postdoctoral Scholars. Housed within the School of Arts & Sciences of the University of Pennsylvania, MindCORE is an interdisciplinary effort to understand human intelligence and behavior.
Applications due December 1, 2026.
mindcore.sas.upenn.edu/post-doctoral-research-fellowship/ -
Lior Pachter
@lpachter
Ummm... www.science.org/content/article/researcher-center-epic-fr…
Or Zamir
@OREAXEAX
Terry's analogy is indeed silly for math, and yet: If human researchers found a vulnerability in the drug trial process, e.g. how to modify proxy biomarkers but not the effect they are meant to predict, they wouldn't knowingly exploit it. Models based on modern optimization would
-
Lior Pachter
@lpachter
What is the maximum quartet distance between two trees?
It's probably best not to talk during salsa but if you're going to chat, how about catching up on the answer to this question?
(based on arxiv.org/abs/2608.03542)This media couldn't be displayed. See it on the original post.Post not available here, view on X
Lior Pachter
@lpachter
All these were produced and rendered on my laptop. I was curious how far this could go... see end of this video (the scene at the end took a few hours on the laptop).
This media couldn't be displayed. See it on the original post.
Lior Pachter
@lpachter
Credits: the salsa video was inspired by x.com/sinabooeshaghi/status/2107966383674077203
The tooling needed for these videos was extensive (see credits). Worthy of special highlight is @3blue1brown's Manim, that will (continue to ) be transformative in terms of facilitating understanding of math.
sina
@sinabooeshaghi
Dude, take a dance class and its pretty clear: exercise, social interaction, attention, touch/partnering, skill building,... all plausible mechanisms to reducing depressive symptoms.
-
Lior Pachter
@lpachter
I’m old enough to remember when mathematicians used to say that citations don’t mean anything in mathematics.
Giulio Tiozzo
@tiozzo_giulio
To highlight the disparity on how subfields have been affected by the OAI drop, here are the #s of citations of all Fields medalists in the repo:
absolutely wild that Thurston has only 1 citation, and Yoccoz has none! -
Lior Pachter
@lpachter
I disagree with Terence Tao's view of mathematics, which he articulated in a talk yesterday at Caltech (see slide below).
Applied math problems are not disjoint from pure math problems. This is not a minor matter.
So no, this is not a "more accurate model". It's embarrassing.Post not available here, view on X
Lior Pachter
@lpachter
I had a coffee with Terence Tao once and he was incredible to talk to. He took a question and within seconds took it to the stratosphere. I couldn't keep up but I love it. It was like talking to a math oracle and it left me wanting more.
I now experience this with AI.
Lior Pachter
@lpachter
Are there serious and urgent problem with AI with respect to mathematics?
Absolutely.
But I'd rather here Tao talk about math. -
Lior Pachter
@lpachter
Mathematicians traffic in ideas. galoisrepresentations.org/2026/10/08/a-brief-digest-of-so…
-
Lior Pachter
@lpachter
Apparently arXiv = GitHub 🤷♂️
Boaz Barak
@boazbaraktcs
Congratulations to Elchanan Mossel
windowsontheory.org/2019/12/30/a-bet-for-the-new-decade/ -
Lior Pachter reposted
sina
@sinabooeshaghi
AI-written science is flooding preprint servers. How can we filter through the noise? Decision language models (like Jev) give us a way forward. New article below. x.com/i/article/2108298508516397056
-
Lior Pachter
@lpachter
Update:
Hundreds of papers published (the package) ✅
The author list is "the consortium" ✅
Raw data not released ✅
Within days errors found in multiple publications ✅
One person claims credit for all the work of the consortium and compares himself to Newton ✅
Lior Pachter
@lpachter
Mathematics has now fully morphed into biology:
Project costs millions $ ✅
Put out paper that nobody has read ✅
Massive paper supplement nobody will read ✅
Advertise with pretty art ✅
Vicious authorship fight ✅
Say you're curing cancer ✅ -
Lior Pachter reposted
Daniel Litt
@littmath
I think you should delete this post—the quote isn’t written by Tao, it’s a guest post on his blog.
-
Lior Pachter
@lpachter
One @OpenAI paper is a 2-approximation for shortest common superstring, a problem relevant to read compression. I tested it and found its construction requires 4.29 million candidate vertices for 1,000 reads, making the algorithm as written impractical at sequencing scale. 1/
Post not available here, view on X
Lior Pachter
@lpachter
This reminded me of the 2015 (conditional) edit distance lower bound of Backurs and Indyk. Back then I wrote about the gap between worst-case asymptotic results and the sequence comparisons biologists actually care about. 3/ liorpachter.wordpress.com/2015/08/14/in-biology-n-does-no…
Lior Pachter
@lpachter
So what's left with these approximation results? Is it, in Hardy's words, merely "aesthetic value"?
Count me unimpressed on these two results, even if they are theoretical breakthroughs. And count me unimpressed with this kind of untethered CS theory in general. 4/4 -
Lior Pachter
@lpachter
I've had a similar experience.
I used to be really into filling out the web forms for submitting a paper to a journal. I'd spend hours doing it. Until one day a student wrote paperpush (github.com/pachterlab/paperpush) and I've permanently lost all interest in submitting to journals.
Michael Baym
@baym
I used to be really into sudoku. I’d spend hours doing them. Until one day on a flight I wrote a sudoku solver (in matlab, don’t judge), and permanently lost all interest in it.
Anyway just thinking about this this morning for no reason whatsoever -
Lior Pachter
@lpachter
Referring to hundreds of solutions to important and difficult mathematics problems, some of them central to entire fields, and many with lean formalizations as "slop" (low quality, mass produced, digital junk) is unserious.
-
Lior Pachter
@lpachter
~80% positive results and ~20% counterexamples.
Interesting and different from some of the predictions I've seen.
(caveat: there's of course ascertainment bias since the denominator is 4,000).
OpenAI
@OpenAI
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math -
Lior Pachter
@lpachter
The primary problem in mathematics right now is not AI, it's sexual harassment. And it's not a new problem. www.facebook.com/noy.soffer/posts/pfbid02XBiktdcidjFMmUnc…
-
Lior Pachter reposted
Gonzalo Cao-Labora
@GonZalocla
This is completely unbelievable and outrageous
We are not talking about accusations here, we are talking about the results of a 3-year long process by a disciplinary committee.
Adiprasito should be out of academia. Otherwise we are putting students at risk -
Lior Pachter
@lpachter
I look forward to seeing how the @NobelPrize foundation selects three agents for the award from among the millions of agents that are used for the Nobel discovery awarded in 2046 (millions of agents who will rely on the entirety of human knowledge built up over centuries).
-
Lior Pachter reposted
Christian Szegedy
@ChrSzegedy
Reading all those recent beautiful essays by Terry Tao, Kevin Buzzard and @wtgowers, I could not resist, but write up my deeply personal perspective on how I perceive and see things:
docs.google.com/document/d/e/2PACX-1vTSh-pyNP3Gi99WMmsinn… -
Lior Pachter reposted
Anshul Kundaje
@anshulkundaje
AI folks who are new to bio should spend some time with wet lab folks not only to understand how experiments, assays, lab work is done but also to learn the basic concept of CONTROLS. 1/
-
Lior Pachter
@lpachter
Crazy the lengths people will go to just to game the STOC 2027 video requirement.
Hassaan Raza
@hassaanraza
The whole point of building Tavus has been simple: talking to a machine should feel as natural as talking to a friend or coworker.
It’s hard to describe all the tiny nuances that make a conversation feel human. The little expressions. Moving around in your chair. Knowing when to speak and when to listen. The dance of it all.
Griffin is by far the closest anyone has come to a model that can capture those nuances. The first time I saw it being used, I had no idea I was watching our model rather than just a normal video call.
I’m so incredibly proud of this team and what they’ve built.
Tavus
@tavus
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).This media couldn't be displayed. See it on the original post. -
Lior Pachter reposted
Nurit Baytch
@NuritBaytch
not just rent. Amsterdam made Holocaust survivors pay the gas bills that Nazis occupying their homes accrued:
www.timesofisrael.com/dutch-student-fought-bureaucrats-fo… -
Lior Pachter
@lpachter
What is a Pachter (other than my surname)?
Pachter is Dutch / German for a person who leases something.
For centuries Jews could not own land in Europe.. hence Pachter as a Jewish surname.
1/Post not available here, view on X
Lior Pachter
@lpachter
So while Amsterdam paid a pittance 70 years late, Leeuwarden paid zilch. Zero.
17/
Lior Pachter
@lpachter
So what is a Pachter? Not just a person who leases.
It's a holocaust survivor returning home to discover that while they had been deported, robbed and murdered... the rent meter was running.
18/18 (in Judaism 18 symbolizes "life" as it corresponds numerically to חי). -
Lior Pachter
@lpachter
We've expanded the checks that paperpush performs prior to submission. We're finding it very useful as a final QC prior to submitting work:
paperpush.readthedocs.io/en/latest/validation-checks.html -
Lior Pachter
@lpachter
This is so pathetic. A company selling AI reviews of papers is reviewed by AI that generates an AI video and post and the company selling AI reviews responds to it as if it was real. 🤦♂️
-
Lior Pachter reposted
Sathvik Redrouthu
@_sathvikr
Read Mathathon's statement on Terence Tao's blog 👇
terrytao.wordpress.com/2026/09/28/joint-statement-about-m…
SAIR
@SAIRfoundation
The @Caltech community and SAIR are redesigning and co-organizing Mathathon: Old Problems, New Proofs.
Explore alternative proofs and clearer explanations, using AI to deepen human understanding.
More on our co-founder Terence Tao’s blog:
terrytao.wordpress.com/2026/09/28/joint-statement-about-m… -
Lior Pachter
@lpachter
Brutal subtweet of the rebranding of the positive-Grassmannian image as an "amplutihedron".
Shriram Krishnamurthi (primary: Bluesky)
@ShriramKMurthi
Oh, clever, rebrand your geometric object to get back into the news cycle.
-
Lior Pachter
@lpachter
I was just reading the same 2021 Nature paper.
Anshul Kundaje
@anshulkundaje
The thing that really bugs me is the confidence they display with sufficiently detailed but also sufficiently vague statements that if u didn't know how to probe, they would sound really fantastic. 4/
-
Lior Pachter
@lpachter
Kill journals?
Paul Novosad
@paulnovosad
Journals should have a zero tolerance policy for AI writing. I know we can think of justified cases, but these will be swamped by an avalanche of slop that will demoralize referees and kill journals entirely.
The bad money drives out the good. -
Lior Pachter reposted
Yun S. Song
@yun_s_song
Please help spread the word! We are recruiting multiple Postdoctoral Fellows as part of the recently launched Bakar Computational Biomedicine Initiative (BCBI) at UC Berkeley and UCSF.
BCBI website: bcbi.berkeley.edu
Apply by Nov 1, 2026: berkeley.infoready4.com
(1/n) -
Lior Pachter
@lpachter
I highly recommned the essay on antedisciplinary science by Sean Eddy written 21 years ago.
journals.plos.org/ploscompbiol/article?id=10.1371/journal…
It seems to me that we are currently experiencing the birth of a new discipline: mathematics engineering. -
Lior Pachter
@lpachter
I wonder if one of the long-standing math problems OpenAI has solved is the leaky pipeline.
-
Lior Pachter
@lpachter
The @usnews rankings are out again and just a reminder that they never show you the relevant object, which is the polytope (rankings with respect to all weights).
arxiv.org/abs/0805.1026 -
Lior Pachter
@lpachter
Subject coverage of the mathematicians comprising the Advisory Group on Mathematics and Artificial Intelligence.
Subject classifications according to the MSC based on arXiv papers.
OpenAI
@OpenAI
We’re working with an independent advisory group of mathematicians to help OpenAI responsibly share advances in AI and mathematics.
The group will advise on how we assess and communicate new mathematical results, uphold academic and professional standards, and build tools that support mathematical research and learning.
Through this work, we want mathematicians to be at the center of shaping how AI supports mathematical understanding and how its benefits reach the wider community.
openai.com/index/advisory-group-on-mathematics-and-ai/ -
Lior Pachter
@lpachter
All this talk about mathematics being "open" is disingenuous. E.g., the NSA discovers math it does not publish.
Differential cryptanalysis was known at the NSA in 1974 but was only published by Shamir after he discovered it independently in 1991.
dl.acm.org/doi/10.1147/rd.383.0243 -
Lior Pachter
@lpachter
Nicholas Polson…
Jeffrey Flier
@jflier
Excellent piece by @UChicago President Paul Alivasatos on how to integrate AI into their revised core curriculum. Sounds good to me!
Wall Street Journal Opinion
@WSJopinion
Teach students to think with, without and about AI. UChicago’s revised core curriculum will require both analog and machine-assisted learning, writes Paul Alivisatos
www.wsj.com/opinion/teach-students-to-think-with-without-… -
Lior Pachter
@lpachter
Maybe parents should just use a random embryo and with the savings purchase an RTX5090 card along with a frontier model subscription to give the baby when it's born.
-
Lior Pachter
@lpachter
In 1999, David Haussler, Richard Karp and I met here to discuss strategies for assembling the human genome. www.berkeleyside.org/2026/09/18/triple-rock-brewery-closi…
-
Lior Pachter
@lpachter
"others" as an author is a great riddle. It could be one person or the entire human population (~8.3 billion). Or it could be a machine (say codex or claude?) Or it could be a dog whose paw was used to press the `return` key.
We'll never know!!
Source: yifzhang.com/blog/KLPO/ -
Lior Pachter
@lpachter
You hire a roofer to replace an old roof. The work is terrible and rushed using surplus tiles and improper nailing. A month later in light rain it leaks.
In a Yelp review must you stick to describing problems with improper nailing or is it ok to also complain about the roofer? -
Lior Pachter
@lpachter
I recommend Serre's "A course in arithmetic".
Elliot Glazer
@ElliotGlazer
The most important field of math is elementary arithmetic.
-
Lior Pachter
@lpachter
Five years ago I wrote about the "amoral nonsense of Orchid's embryo selection". Since then, the amoral nonsense has multiplied with companies doubling and tripling down, new companies launched, and an overall booming industry.
Super depressing.
liorpachter.wordpress.com/2021/04/12/the-amoral-nonsense-…
Lior Pachter
@lpachter
"Humanity can too"? Their IQ analysis is massively inflated by an assumption their own source told them not to make! 1/
Kian Sadeghi
@KianSadeghi5
AI is rapidly getting smarter.
Now, humanity can too.
Today, @nucleusgenomics is announcing Vitruvian, our newest set of genetic optimization models.
Vitruvian’s intelligence model can optimize embryo DNA for 14 IQ points — nearly a standard deviation.
The models were trained on 1,000,000+ people, validated across 40,000+ siblings, and used more than 7 million genetic markers.
In Superintelligence, Nick Bostrom proposed genetic optimization as a key way for humanity to keep pace with rapidly advancing AI.
Bostrom’s vision is no longer theoretical.
Genetic optimization, like AI, has followed a scaling law: as datasets have grown, so have model capabilities. This trend will continue.
Parents across the world now have the choice to substantially increase their child’s intelligence.
And the next generation can choose to do the same.
AI is no longer the only intelligence that will compound.
Humanity can now direct its own evolution.This media couldn't be displayed. See it on the original post. -
Lior Pachter reposted
Anshul Kundaje
@anshulkundaje
I wonder if cherry picking results, exaggerating results, hyping, not being honest about limitations, not sharing data or code were the best practices? 6/
-
Lior Pachter
@lpachter
"Humanity can too"? Their IQ analysis is massively inflated by an assumption their own source told them not to make! 1/
Kian Sadeghi
@KianSadeghi5
AI is rapidly getting smarter.
Now, humanity can too.
Today, @nucleusgenomics is announcing Vitruvian, our newest set of genetic optimization models.
Vitruvian’s intelligence model can optimize embryo DNA for 14 IQ points — nearly a standard deviation.
The models were trained on 1,000,000+ people, validated across 40,000+ siblings, and used more than 7 million genetic markers.
In Superintelligence, Nick Bostrom proposed genetic optimization as a key way for humanity to keep pace with rapidly advancing AI.
Bostrom’s vision is no longer theoretical.
Genetic optimization, like AI, has followed a scaling law: as datasets have grown, so have model capabilities. This trend will continue.
Parents across the world now have the choice to substantially increase their child’s intelligence.
And the next generation can choose to do the same.
AI is no longer the only intelligence that will compound.
Humanity can now direct its own evolution.This media couldn't be displayed. See it on the original post.
Lior Pachter
@lpachter
Nucleus is claiming that embryo screening can shift IQ by 10.8 points. But that number depends on the assumption of no assortative mating. That is false for cognitive traits (and well known to be false!) In fact the 10.8 is based on a paper explicitly making that point! 2/
Post not available here, view on X
Lior Pachter
@lpachter
Apply an assortative mating correction using published estimates and the IQ spread drops by ~20%. But that's just a guess.. If Nucleus has 20,000 sub pairs they could produce a more exact estimate. So why didn't they? 5/
Lior Pachter
@lpachter
The Nucleus ad is littered with other major problems. Height population R^2 = within-family R^2 to 3 decimal digits?! Mismatched cohorts. 14 = 7? I mean is this really truthful advertising? 6/6
-
Lior Pachter
@lpachter
TIL that ICLR was conceived as a 6-7 joke.
Denny Zhou
@denny_zhou
ICLR 2027 has received more submissions than all previous years (2013–2026) combined
-
Lior Pachter reposted
Zane Koch
@zanehkoch
Science – debatably the #1 journal in the world – published a paper on a "virtual biotech" AI system this week
but the authors never even test if it works well!
all the evidence for its performance is 3 anecdotes
the paper has no benchmark scores, no baselines, no replicates -
Lior Pachter
@lpachter
Individuals should be limited to submitting a single paper in their entire career.
This will reduce the reviewing load at conferences and journals, p-values will not need to be corrected for testing performed in multiple papers, and publication charges will be manageable. -
Lior Pachter
@lpachter
This is interesting. www.nature.com/articles/s41593-026-02433-7
-
Lior Pachter
@lpachter
LOL
Parsa Hosseini
@paahrsa
45k submissions for @iclr_conf so far 💀💀
-
Lior Pachter
@lpachter
Rushing to make the ICLR deadline?
Paperpush will save time on the upload github.com/pachterlab/paperpush -
Lior Pachter
@lpachter
Math departments ca. 2026
-
Lior Pachter
@lpachter
Everybody knows it's the discovery of an algorithm for the halting problem.
𝖬𝖺𝗁𝖽𝗂 𝖢𝗁. (bluesky:@mahdi.ch)
@mahdi_tcs_
Since people are randomly predicting: My gut feeling says that, other than proving UGC, @OpenAI may have proven than the Strong ETH is false! Maybe log-rank (false!), Graph Iso in P, & strong polytime LP come to mind too. Probably NOT MatrixMul ω=2 or Quantum PCP or PRG for L. 🤷♂️
-
Lior Pachter
@lpachter
LOL
"There are also areas where I believe it is explicitly permitted to use LLMs. They can provide translation services" -
Lior Pachter
@lpachter
• Hoards solutions to math problems and doesn't share
• When finally sharing solutions to math problems deliberately hides how the solutions were arrived at
Clamorem Boeotorum vita et pauca sed matura. -
Lior Pachter
@lpachter
Beautiful paper on how to tease apart cis vs. trans contributions to differences in gene expression. Work of Ingileif Hallgrímsdóttir and Maria Carilli just out in Genetics academic.oup.com/genetics/advance-article/doi/10.1093/gen…
1/🧵Post not available here, view on X
Lior Pachter
@lpachter
The first observation is that there is a simple linear transformation that simplifies interpretation of the coordinates. 3/
Lior Pachter
@lpachter
Testing with a GLM can allow for effective utilization of replicates. We reanalyze very cool data from the Nachman lab to make the case. But there is a meta point here... 4/
Post not available here, view on X
Lior Pachter
@lpachter
The method has already been used by others; see e.g. elifesciences.org/reviewed-preprints/110575 who studied methylation. 10/
Lior Pachter
@lpachter
I was mostly a fly on the wall for this project but it's been a ton of fun watching the math and stats come together in this genetics setting. 11/11
-
Lior Pachter
@lpachter
This is a thoughtful essay. The call for a rigorous defense is one I think needs to become universal in all fields, not just math.
I will just add that the suggestions (internal and social-relational) apply at all level (including high school).
Daniel Litt
@littmath
I've written an essay on how I think the mathematics profession should adapt to highly capable AI systems. It's hosted here on "Proofs and Prompts": proofsandprompts.com/2026/09/14/a-beginning-for-mathematics/
though you should also feel free to complain/comment on my website here: www.daniellitt.com/blog/2026/9/13/a-beginning-for-mathema… -
Lior Pachter reposted
Hani Goodarzi
@genophoria
We got really good at DNA sequencing a long time ago. Ever since, any problem that can be turned into a sequencing problem can be magically solved, giving rise to *-seq methods that form the backbone of modern genomics.
This team at Escalante has done this for protein binding kinetics. A million interactions in one tube, one overnight reaction, one sequencing run. Worth the read... blog.escalante.bio/how-to-sequence-fundamental-biophysics -
Lior Pachter
@lpachter
Work is proceeding on the Hodge conjecture.
-
Lior Pachter
@lpachter
Actually there is now only one stage. Push the button (available only to employees of openAI and Anthropic. Usage limits do not apply. Discoveries announced on social media).
Itai Yanai
@ItaiYanai
The 3 stages of a discovery: first they'll say you're wrong, then they'll say it’s not important, finally they'll credit someone else.
-
Lior Pachter
@lpachter
The transformation of mathematics that has taken place over the past 8 months is astonishing and historic. Incredible experience to live through this.
Math Files
@Math_files
Mathematician Joel David Hamkins has sharply criticized AI models for mathematical work, calling their usefulness “essentially zero.”
He pointed out that these models often state wrong answers with confidence and resist correction, making them unreliable.
Hamkins remarked that if a person behaved this way, he would simply stop engaging with them. -
Lior Pachter reposted
Mathurin Dorel
@MathSRIsh
This take is absolutely wrong and a major reason lay people overestimate AlphaFold's capabilities.
Biologist have carefully examined the capabilities of successful AlphaFold versions and found them useful in some case and severely lacking for most out of distribution proteins.
Tristan Farmer
@001TMF
Biologists: "It folded a protein? Awesome. Don't care how the black box works, let's cure diseases."
Mathematicians: "It hallucinated a single lemma in step 43. Burn it to the ground." -
Lior Pachter reposted
Rachel Thomas
@math_rachel
"The goals of AI companies do not seem to be aligned with 'the primary goal of conceptual understanding & insight', but the goals of the math community do not seem aligned with this goal either. Sadly students & ideas in math have not been 'nurtured with great care.' - @lpachter
-
Lior Pachter
@lpachter
I've considered signing the "Severe Misalignment of AI in Mathematics" declaration because I care about the future of mathematics. But I'm not going to. Here's why
liorpachter.wordpress.com/2026/09/12/align-ai-and-mathema… -
Lior Pachter reposted
Mohit Kumar Jolly
@mkjolly15
Humbled and honoured to be elected as a Fellow of the National Academy of Sciences, India (NASI). Grateful to my nominators, mentors, collaborators, past & present members of our research group, and funding agencies.
@ANRFIndia @hansa_para92520 @ICMRDELHI @DBTIndia @EduMinOfIndia
BE, IISc
@BE_IISc
Heartiest congratulations to Prof. @mkjolly15 on being elected as a Fellow of the National Academy of Sciences, India (NASI)! 🎉
A proud moment for the @BE_IISc, and a recognition of his outstanding contributions to cancer research.
@nasi_academy
@iiscbangalore -
Lior Pachter
@lpachter
Thinking today of Danny Lewin ז״ל, my classmate in grad school who was the first person murdered in the terror attack of 9/11.
My last memory of him is vivid in my mind’s eye. He’s leaning back in his chair, feet on the desk, hands behind his head, his laughter filling the room. -
Lior Pachter
@lpachter
Mathematics has now fully morphed into biology:
Project costs millions $ ✅
Put out paper that nobody has read ✅
Massive paper supplement nobody will read ✅
Advertise with pretty art ✅
Vicious authorship fight ✅
Say you're curing cancer ✅ -
Lior Pachter
@lpachter
The OpenAI Navier-Stokes paper is 166 pages. Work on the problem began one week ago.
That means the paper must have been written (by machine) in something like a day or two. I'm curious if even a single person has read it yet? -
Lior Pachter reposted
Rasmus Nielsen
@ras_nielsen
Staff Scientist position at University of Copenhagen. Come join our work on ancient environmental DNA! candidate.hr-manager.net/ApplicationInit.aspx?cid=3010&Pr…
-
Lior Pachter
@lpachter
My 2¢:
x² + 10⁻⁴² = 0 has an “approximate real solution” at x = 0 with a residual of 10⁻⁴². It has no real solution.
Small residuals are impressive. Turning them into a disproof requires... math (an existence argument). -
Lior Pachter
@lpachter
I look forward to reading the New Yorker article.
-
Lior Pachter reposted
Hafsteinn
@hafsteinn
🎓 Laust starf lektors í tölvunarfræði eða hugbúnaðarverkfræði við Háskóla Íslands. Umsóknarfrestur 2. október.
🎓 Open position: Assistant Professor in Computer Science or Software Engineering at the University of Iceland. Deadline 2 October.
hi.is/lausstorf -
Lior Pachter
@lpachter
The MoTrPAC consortium has been studying the molecular effects of exercise. In a new paper with Conrad Oakes, we take a careful look at their rat dataset, leading to discoveries that highlight the importance of proper bioinformatics & stats. 1/ link.springer.com/article/10.1186/s12863-026-01440-8
Post not available here, view on X
Lior Pachter
@lpachter
We were able to find a sick rat using @NeuroLuebbert's viral detection method (see www.nature.com/articles/s41587-025-02614-y)! This matters because health is an important covariate. the Tigit gene can appear differential due to exercise, when in fact it's immune response due to sickness. 5/
Lior Pachter
@lpachter
tl;dr #methodsmatter. Our results are fully reproducible via @GoogleColab github.com/pachterlab/OP_2025
-
Lior Pachter
@lpachter
This is an extremely high bar for what one should consider AGI.
I've known many mathematicians who are considered "generally intelligent" yet can't order pizza (even though they can play chess). Some can get pizza ordered, but can't play chess. Paul Erdős could do neither.
Gary Marcus
@GaryMarcus
the problem with claiming AGI is that it means GENERAL intelligence.
which means you have to do more or less everything cognitive, from ordering pizza to playing chess.
the whole reason the term was invented was that older systems were specialists (e.g., chess) but not general.
that is still the case. period.
Trini Zone
@TriniZone
I mean agents still can't even order me a pizza on demand yet 😂. But I guess every business on the planet has to market their product.
-
Lior Pachter
@lpachter
I just heard a frightening primal scream.
Sounded like the words “manifold destiny” in a thick Italian accent. Could it have been Tartaglia?
Другая планета
@nihilunbounded
Much as I see the heavy irony in Buzzard getting undermined by a monster of his own making, I agree with this completely. Commonly understood social norm in math to not scoop someone on their research project. Increasingly clear that Anthropic gives less fucks about math than OAI
-
Lior Pachter
@lpachter
“The honest claim” is a new phrase showing up in @biorxivpreprint, frequently without the honest explanation that the text was written by AI.
-
Lior Pachter
@lpachter
To wish that mathematics becomes like the human genome project is to wish that mathematicians not understand mathematics. Because one thing the human genome is most definitely not is "fully classified".
Jared Duker Lichtman
@jdlichtman
We are now poised to:
Formalize all known Human Mathematics.
This is the modern equivalent of the Human Genome Project. or AlphaFold more recently, whereby all data of its kind is fully classified.
With sufficient funding and compute/tokens, a coordinated effort among academia, frontier labs, philanthropic organizations, and the federal government can formalize all Human Mathematics in the next year. -
Lior Pachter
@lpachter
Proof that 2035 < 2032.
All that remains is to formalize this in Lean 4. -
Lior Pachter reposted
John B. Holbein
@JohnHolbein1
You can literally see how Dan Ariely* manually changed points in his 2002 paper to produce the results he wanted.
This is truly remarkable levels of data fraud from one of the world's most famous social scientists.
*I'm guessing Ariely will land on the Gino dog-ate-my-homework defense and say that a mysterious RA who was not a coauthor and thus had very little incentive to selectively manipulate the data did it. I doubt it. -
Lior Pachter
@lpachter
So a bunch of companies are expecting mathematicians to spend their "time to verify these results" for free..
After expecting talented undergrads to prompt the models for free..
Dressing advertising as academic collaboration..
"Sponsored hackathon" is just a euphemism for grift.
Sathvik Redrouthu
@_sathvikr
A few @Caltech friends and I are hosting the first math hackathon with frontier labs/startups.
We're giving $1M+ of compute to 100 teams to solve the hardest open problems -- competitors range from IMO golds to frontier/neolab folks to math PhDs/profs.
DM me if your company wants to be involved :) we are launching soon. -
Lior Pachter reposted
Richard Sever
@cshperspectives
‘Unrefereed, unindexed and unread’ was the joke I always made about supp data.
AI will change that…
Lior Pachter
@lpachter
An underappreciated impact of AI on biology is that an enormous, I mean ENORMOUS, number of skeletons buried in genomics papers are all about to burst out of their Supplementary Materials.
-
Lior Pachter
@lpachter
An underappreciated impact of AI on biology is that an enormous, I mean ENORMOUS, number of skeletons buried in genomics papers are all about to burst out of their Supplementary Materials.
-
Lior Pachter reposted
Richard McElreath 🐈⬛
@rlmcelreath
Someone emailed me to ask if my lectures are online anywhere, so yes here is the link to most recent lectures. These lectures are free and free of Cyclospora
Richard McElreath 🐈⬛
@rlmcelreath
Statistical Rethinking 2026 is done: 20 new lectures emphasizing logical & critical statistical workflow, from basics of probability to causal inference to reliable computation to sensitivity. It's all free, made just for you. Lecture list & links: github.com/rmcelreath/stat_rethinking_2026?tab=readme-ov-…
-
Lior Pachter
@lpachter
The @iclr_conf 2027 call for papers is now live. @Josephmrich has added ICLR to paperpush so you can now submit your 20 papers (the quota this year...) with a single click! github.com/pachterlab/paperpush
ICLR
@iclr_conf
The ICLR 2027 CfP is up! iclr.cc/Conferences/2027/CallForPapers
-
Lior Pachter
@lpachter
For users of edgePython: an update (not yet in a release but in main) providing a substantial speedup thanks to Peter Haverty. github.com/pachterlab/edgePython
-
Lior Pachter
@lpachter
Thanks to a pull request by @fisherastro span now includes an (adversarial via LLM with a well thought out protocol) audit for checking that lean statements correctly align with theorem statements in the corresponding paper. github.com/pachterlab/span
-
Lior Pachter
@lpachter
Never attribute to AI slop what can be attributed to incompetence.
-
Lior Pachter reposted
Uthsav Chitra
@uthsavc
Do spatial transcriptomics foundation models (SFMs) actually learn biologically meaningful spatial patterns? 🧬
Excited to share our new work assessing if current spatial foundation models learn multi-scale spatial variation in gene expression (spoiler: mostly no) -
Lior Pachter reposted
sina
@sinabooeshaghi
"Hey claude write me tweet that explains the Moderna/Merck result. Explain the pipeline, the ML, the trial design, and what it validates. Make no mistakes"
Bo Wang
@BoWang87
Moderna/Merck just ran a 1,137-patient Phase 3 trial where every single dose was unique to that patient's tumor. It worked.
The pipeline: surgical resection → whole exome + RNA sequencing → ML neoantigen ranking → mRNA encoding up to 34 patient-specific targets → manufactured and shipped in 8 weeks. One drug, different sequence for every patient.
The ML step is worth to note: the algorithm ingests WES + RNA-seq to identify somatic mutations, then predicts which of those will actually be immunogenic, ie, displayed on tumor cell surface and trigger a T-cell response. It's designed to keep learning from accumulated clinical and immunogenicity data across patients, not just per-patient.
INTerpath-001 (Stage IIB-IV resected melanoma, 2:1 randomized): combination with pembrolizumab beat Keytruda alone on both primary (RFS) and key secondary (DMFS) at interim. Phase 2b at ASCO 2026 showed 49% reduction in recurrence/death, 59% in distant metastasis/death at 5 years. Phase 3 confirmed both.
What this validates:
— tumor-specific neoantigen prediction by ML works in a blinded trial at scale
— 8-week personalized mRNA manufacturing is operationally real
— effect is additive on PD-1 blockade, not redundant
This is first positive Ph3 for individualized neoantigen therapy. First positive Ph3 for any mRNA cancer therapeutic.
What a great time to live in!! This is the best time for AI & biotech! -
Lior Pachter reposted
Chang M. Yun
@chang_m_yun
Just read the griftiest thing. Seems like all it takes to be AIxBio is
1/Copy open source models, slightly tune, rename {company}-{modality}
2/Make pretty UI(video only)+API(code block only) wrapper
3/Show it "works" in K562+HepG2 which have experiments so is just retrieval
Done! -
Lior Pachter reposted
Vince Vatter
@VinceVatter
New Zeilberger opinion dropped
-
Lior Pachter reposted
Wayne Riekhof
@wayneriekhof
Always astounds me when prepping for my fall course on microbial physiology that this one little segment, about 100kb, or 2% of the E. coli genome, just about encompasses the whole course. The Dcw operon alone, the whole third row, that is... good Lord! Amazing.🤯
-
Lior Pachter
@lpachter
These are the grade distributions in intro calculus at UC Berkeley 2024 and 2025.
You can see the massive difference due to no longer considering the SAT in admissions. From almost a quarter A+ in 2024... to Cs and Ds!!
Oh wait a minute... when exactly was the SAT dropped? 🤔Post not available here, view on X
Lior Pachter
@lpachter
I first wrote about this in 2013 (!) when I noticed that in a single semester two different instructor assigned wildly different grades... in the same intro calculus class!! liorpachter.wordpress.com/2013/12/23/time-to-end-letter-g…
Lior Pachter
@lpachter
Of course it may be the case that the preparation of students has changed over the years. And it may be the case that the Reading and Writing sections of the SAT are valuable for admissions.
But one thing is for sure: if you get in choose your instructor carefully! -
Lior Pachter
@lpachter
Who's going to tell him about paperpush? github.com/pachterlab/paperpush
Auyon Siddiq
@auyonymous
There's a chaired, full professor at UChicago who has put out *200* papers in 2026 alone. And at least one has been submitted to a journal (so potentially many more).
They don't all cite each other though -- rookie mistake. -
Lior Pachter reposted
Michael 英泉 Eisen
@mbeisen
I told them they should have called for reinstating the verbal SAT too....
The Daily Californian
@dailycal
SCOOP: UC Berkeley math professor Zvezdelina Stankova admitted to using an AI tool in an op-ed urging the UC system to re-adopt SAT and ACT requirements in admissions.
The article, published in the SF Standard, was flagged as 33% AI-generated or AI-assisted by detector Pangram.
A similar result was found in a June open letter from STEM faculty advocating for standardized testing. The letter was signed by thousands of academics, including five Nobel laureates, and was partially written by Stankova.
Read our coverage at the link below: www.dailycal.org/news/campus/uc-berkeley-professor-disclo… -
Lior Pachter
@lpachter
Shocking that calculus students nowadays are so unprepared that they think one can apply the distributive law to log functions to get log(1+2+3)=log(1)+log(2)+log(3).
sfstandard.com/opinion/2026/08/15/uc-berkeley-sat-test-bl… -
Lior Pachter
@lpachter
We don’t even know what all the human diseases are.
-
Lior Pachter reposted
-
Lior Pachter reposted
Gavin Crooks
@gavincrooks
Physics is going to be as cooked/cooking as math. I fed Claude an open problem in stochastic thermodynamics of the kind I'd suggest to a mathematically inclined grad student. And over a few days of back and forth, it did months of work and closed the whole problem class.
-
Lior Pachter
@lpachter
"The fertility decline is nearly global."
In Africa there is a decline, yes, but the estimate is for fertility rates to reach replacement only around 2100. In the meantime, the population is growing rapidly (currently ~4 births per woman) but somehow Africans are always ignored.
Daniel Lemire
@lemire
Lance Fortnow is a well-known computer scientist. He just lost his tenured position at Illinois Tech in a wave of cuts driven by a financial crunch. I have his book on P vs NP on my shelf.
Here is how he described it:
« Illinois Tech got hit hard by a large drop in foreign graduate enrollment due to changing visa requirements, general anti-US sentiment, more opportunities in other countries and a weaker job market for graduating Masters students partly due to artificial intelligence. Universities face challenges beyond international students including Baumol's disease, administrative bloat to meet expanding regulations, the demographic cliff, reduced grant funding, and less support of universities by the public and both political parties, and AI changing how and why we teach. Illinois Tech is one of the first tech research schools to eliminate tenured roles, but it won't be the last. »
His school lost about 1,200 international graduate students (Chicago Sun-Times). Roughly 60% of its graduate students came from abroad.
A similar contraction is underway in Canada. It has become harder for international graduate students to enter, and several universities have already cut positions.
Two further points matter.
1. New administrative roles appear constantly. Fortnow attributes this to expanding regulations. That is only part of the story. Universities themselves chose to expand the administrative apparatus while holding faculty and student numbers relatively flat. Not every new administrative position was forced by external rules.
2. Collapsing birth rates mean fewer young people to educate. This is mathematical and not reversing anytime soon. Importing students from elsewhere only shifts the shortage to their home countries. The fertility decline is nearly global.
Will we see more cuts like Fortnow’s? Universities in the US and Canada should not expect long-term growth under current conditions. On average they will shrink. The administrative overhead accumulated over the past decades makes the adjustment painful. Institutions face a choice: cut red tape and administrative staff, or cut people like Fortnow.
Thus far, it has been a slow-moving, almost impossible to notice process. Increasing the number of foreign students has effectively hidden the domestic issue (slightly fewer students on average each year, and university attendance not rising). But this is unlikely to prove a lasting solution everywhere. -
Lior Pachter reposted
Yadin Dudai
@yadindudai
אדם דגול הלך לעולמו. יחיד ומיוחד. יהי זיכרו ברוך.
-
Lior Pachter reposted
eLife - the journal
@eLife
Today, we’re pleased to share an overview of the partnerships we’re involved in to harness AI for improving research engagement. elifesciences.org/inside-elife/6ad10e4e/elife-latest-expl…
-
Lior Pachter
@lpachter
In 1999 I applied for a handful of jobs in math departments. Turns out someone recorded one of the interviews I had.
I had just been asked whether I'm really a "mathematician".
Hassaan Raza
@hassaanraza
Warning: Dropping attention on video diffusion models may cause nightmares
-
Lior Pachter
@lpachter
Biologists are cooked.
levent
@__alpoge__
+++-+-+-+-+--+--+++-++----++--+----+-----+++---++-+---+++++++-+--+-+-+--++-+--+--+----+-+-+-+-++-++---+--++++--++-++++-+++++--+---++-+---+++++++++--+-+-+--++-+--+--+-----+++-+++-++-+-++++-++-+--+--+------++--+-++-+-+---+--+-+-+-+--++---+++++-----+++++++---+---+--+-+--+-+--+-++-++-++++++-+-++-++-+-+---+--+-+-+-+--++---+++++-----+++----+-+-++---+-++----+---+++++--+---+--++-+++++++-+--+---+----++-+++----++---+++---++++-+-+--+++-+--++++-+++-----++-+++-+++-++++++++-+--+---+----++-+++----++---+++---+++---+-+--++----+--+-+++-+--++------+++--++++++-+--++--+-++---+--++---+---++--+-+----+++-+-++--++++-++-+---+-++--++++++---++++++++-+--++--+-+++--+--++---+---++--+-+-+--+++--+---++++-+-++-+---+-+-+-----+--++-+----+-+-+----+-----++-+++++-+++--+--++++++-++--++---++-+++----+-+--+--++-+-+-+++++-++--+-+++-+-+-+----+-----++-+-+++-+++--+--++++++-++-+++---++++---+++++-++--+--+++-+++--+--+-++-+++--+++-+++--+-+--++++-----+-+-+--+--+---+-++---+++----+++-----+--+++++---+---++-++-+--+-++--+++-+++--+-+--++++-----+-+-+--+--+---+-+++-------+-+-++--++++---+--+-++-+--+---+-++-+-+++++-++--+----++++----+++---++--++-+---+-+--+++++++-+-+--++----+++-++-+--+-++-+++-+--+++-++++-++--+----++++----+++---++--++-+---+-+-------------+-+++-+++---++--++--+-+++--+-+-++---+----++++--+-+-++++-++---+--+----++-++-++++++++++++++-+---+---+++--++--++-+---++-+-+-+-+-+----++++--+-+-++++-++---+--+----++-++-++----++-++++--+++++--+--+--+--+++-----++--+-+-+++-+++-+++-+-+++-+++-+++++-++++---+++---+--+--++++--+++++--+---+++-+--+-----+-+++++++------+++-+--++-+++--++--+++-+-+++-++-+++-+----++-------+--+-+-++-+-+--+--+-+--+-+-+-++-+-+---+++++-+-++-++----++--++++-+--++-+---+-++--+-+----+------+--+-+-+--+---+++++-+++-++++---+-++-++--+-++-+---+++-+-+++-+-+--+++-+-----++-+++--+--++++-++-+-++++--+--++---+++++---+---++++-++----+--+-+-++++++-+++-++++--+---+--++--+-+-++-+++-+++---++---++----++------+--+--++++-+++-++++--+++++-+-+-++++-+-+-++--+++++-----+++-+++++-+---+--+++++-+---++------++-+++-+++-++---++-+--+--+-++++--+-+++-+++-+---++++---++--+-++-+--++-+------+-++++-+--+-+---++---+-++++-++++-+--++++---++---+++--+--++++-++-++++++-+---+-----+++--++----+----+-+-+----++++++---+-+---+++++-++---+-+--++--+-++-+++-+++-+++++-+-+-+---++-+++-+++-++---++-+-+--+--+--+-+-++-++++-++--+++++-+--+++++-+--+-++-+++---++-+----+-++--+---+----+----++--++-+++-+-++-++-++++--+++-+-+-+-+---++++++-----+++-+--+-+--++---++++-+-+-+-++-++-+--++++-++++++-++--+++-+++---+--+++++++++-++++----+++----++----+-++++--+--+--+-++--+-++--++---+-+--+++-+--++-++--++-+-+----+--++-++++-+--++--+-+--+-+++--+-++---+-+-+-+++++++--++-+-++++-+--+-+-----++--++---+------+-++-+-+-+--+++-++--+-++++++-+++++----++++++++-----+---+-++---+++++++++++-++++-+---+-+++--+-++-++-+-+-+++++-----++-++-+-++-+--++----+--++--+-++--+-+++-+-+-+-+--++++---+--++------++-++++-++----++-+-+--++-+++--+++--+++---+++--+-++++-++---+++---+-+--+-++-++---+++-++----+-+--+++++++++-+-+--+-----+--+++++---+-++++-++++-+---+++++-+---+--+--++--+---------++--+--++++-+--+++-+-+---+----++---+-+++--+-+++-+++-+----++-++-++++--+++-++-+++-++++++---++-+++---+---++++--+-+-+-+++---++--++++---+-++-+-+-+++++++--++-+--+-+-++++++---++----++---+++---+-+++++-----+-+-++-----+-++++-++---+-+-+++++--++++++++--+-+-+--++-+++------+++-++-+++----+---+-+++++-+--+++--+--++-+--+++--++-++++-+++-+++-+----+--+----+-+----++--+++++--++-+-----+--+--++-+-+-+--+-+-----++----++-+-++-----+--+-++-++++-++++-++-+-++-+-+-+-+++-++--+--++--++-+++---+--+-+---++++--+-++-++---+-+-+++-+---++++-+---++-+++++++-++++--+-----+--+++-++--+-+++++++-+-+--+++----+----++++--+++-----++++-+++-+-+---+++++++-++++-----+++-+--+++--++-+-++---+--++--+---++-++---+---+-++-++-++-+-+-++-+-++++++-++++--++--+-++-+++-+---++++--+-----+++-++--+-+-+-++++-++-+--+-+---+-+-++---+-+--+----++--+-+-+-+++--+-++-++++-+---+++---+---++-+++--+--++-+--+----+++-+-++-+++--+++++-+++--++-+++--++--+----+++++-++-++++++-+++++++-+--+++-++++---+-+---+--+--++++-+-+----++-+---++-+--++-+---+-++++-+--+++-+-+++++---++-+--++-+-+-+++++-------+++-++--+-+----++++--+--+-+---+++++--+-+++++-+---+----+---+++-++--++----+-++--+-+++++--++-++---+++++--++-+--+++-+-+++-+-+-++-++++-++--+--+---++--+--+++-+-++-++--+---+----++--+++--++---++-+-+++----+++++-++-++++++-++-++-+++-+-+-+----+-+----+++-+++--+++++++--++-+++++-+-++++----+--+-+-+++++++++---+---++--+-+++--+-+--+-+++-+-+--++-+-------+++++-+--+-+--+-+-++------++++--++-----+-+--+----++----++++---+-+-----+-+----+--+-++-+-+--+++-++++--++++--+-+----++--++-+-++++-++----------+-++-+-+-+-+-+-+----+----++--+--+++-++--++-++--++++--+-+--+-----++-+++-+-+++-++--+++--+-----+++-+---++-----++-+-+++-++----++-++-+-+-------++--++--+-++-+--+-+++++-+--+++++---+--++++++---++--+-+++-++++--+++-+--++--+++++++++--++-+--+---+-+-++++-++-+--+---+-+-++++-++-+--+---+-+-+++-+-+--++-+-+----+--+++-+--+--+++++---+++--++++++-+---+-+--+-+++++++++-+---+-+++---++-+-+--+++--++--+--+---++-----+--+++-+--+--+-++--++-+++-+-++-+--+----+++++-++-++--++------------+-++++-++-+++-++--++--+---+---+++--+++-++-+++--++-++-++-+----+-++-+++-----+-++++++--+-+++-+---++-++-+-+---++---+-++--+---+++-++---+-+---++---+++++---+-++-++++--+++-+-+-+++++++++---++++++---++-----+++--+-+-+------++-+---+-+++--++-+-++-+-+-++++----++-+-+++---++--+++--+--+---+-+-+-+--++--+--+-++++++--+-+--+--+----++-++++++-+--+--++-++--+-+---++-+--+-----+-++++----++-+++-+-+++---++-+-++-+++-+--+++-----++-++----+++--+++-+++++++-++++-+---++------++--+++----++-+++++---+-+-+-+-++-+-++-+-++--++----+--++-+++-+---++-+-++-+---+----+++-+++-++-+++--++--+-+---+-+-+++++-+-+++--+++++--+++-----+-++------+-++-++---++-+-++++++--+-++-++----+-+-++++--+--++-+--+-+-++++-++--++----+-+-+++++-++++--+++-++++++-+++-+---+-++--+---+++--+-+++-+---+----++-++---+++-+-++---+++-+-+-+-++-++-+--++--++-++-+---+--++++-----+++-+++--+--++-----+-+--+++++-++-+-+--++++++--+--+-++++++--++-+----+------++++----+++++--++-++++-++++-+--+++--+-+++-+-++++-++-----+++-+++-+-+--++-+--++++--+----+-+-+-++-+-+---+---+-++--+---++++-++-++----++-+++--+++++-+++-+-+---+-+----+-++--+++--+-++--+-+++++++++-+---+++++-----+++----+++--+-+-++--+++--+--++-++++-+++++--+---+-+----+--++++--++-+++-+++++-++++++--++---+-+++-++--+---++--++-+--+-++---+----++-+---++--++--+-+-+++--+---+++----+-+-++--++--++-++-+---+-+-++--+-+-+----+---+--+-+-++-++-+--+--++-+-+-+-+-++--+--+--+---+++-+--+-+++--+-+-+-+--+-++--+-++-+--+---+++---+-+-+-++-+--++-+-+---++-+--++-+-++--++-++-+-++--+-+++--+-++---+--+-++--+-++-++-+--+---++--+-+-++-+--+++---++--++-+--++-+++++-++-+++-+++----+-++---++-+--+--+++-+-+-+++++---+-++-+--+++++--+-+-+--++-++-+--+-+-+---+++-+---+-++--++-+-++--+--+-+-+++---++-+---+----+--+--++-+--++--+-++-++--+---++-+--++--+++-+--+------++--+-+-+--+-++-+--++--++--+++---+-+-+++----++--+-+-+---++-+--+-++++-+--++++-+------++++++--++---+-+-++---++-+++--+-----+---++++++++++--++---+--++-+-+-++-+-+--+-+-+-----+-+--+++-+++---++---+++++---+--+++-+-+-++---++---++-++++---+++++---+++---+++---+-+---+++--+--+++++-+++---+---++++---+++---+-+-++++++-+---++++-++++----+++--+++------+++++++++++++++++++++++-+-++-+-------++--++-----+-++-+-----++++++++----++++--------+-+-+-+-----++--++-------+-+-+-+++++-----++---++--++-+-++--+-+-++--+++--++---+-++--+-++-+--++-+-+--++--+++----++-+-+-++-+-+-+--+++----+++-+--++--++--++-+-+--++---++++---++--++-+--++++-++-+---++-----+++-----+-+++-++-+--++--++--+++--+-++--++-++-+---+--+-+-++++---+--+++-++-+---+++----+++++++-+-+--+----+++--+--+----+++-+-++---+-++-+++--+--+-+--++-----+-++--+++++++-+-+++-+---+--+-+--+--+---+-+++++-+-+-+------------+---+++-+-+-+++---+-+-+---+++-+-+-+++---+---+---+++-+-++++++++++--+---+++-+-++++++++++--+---+++-+--+---+++-+--+---+++-+--+---+++-+-+-+++---+-+----------++-+++---+--+++++++++++-+---+++-+--+---+++-+-+-+++---+--+-+++---+--+-+++---+-+-----------+-+++---+--++++++++++--+---+++-+-+-+++---+-++-+++---+---+---+++-+-+-+++---+-+----------++-+++---+---+---+++-+--+---+++-++-+---+++-++-+---+++-++-+---+++-+++-+++---+-+-+---+++-+++-+++---+-++-+++---+--+-+++---+-+-+---+++-+++-+++---+--+-+++---+-++-+++---+-+-+---+++-+++-+++---+-++-+++---+-++-+++---+-+-+---+++-+-++++++++++--+---+++-++-+---+++-+++-+++---+-+----------++-+++---+-+-+---+++-+-+-+++---+-++-+++---+-+-+---+++-+++-+++---+-+-+---+++-++-+---+++-+--+---+++-+-+-+++---+--+-+++---+--+-+++---+---+---+++-+--+---+++-+-+-+++---+---+---+++-+-+-+++---+--+-+++---+--+-+++---+--+-+++---+--+-+++---+---+---+++-+-+-+++---+---+---+++-+-+-+++---+---+---+++-+-+-+++---+--+-+++---+---+---+++-+--+---+++-+-+-+++---+---+---+++-+--+---+++-+--+---+++-+--+---+++-+--+---+++-+-+-+++---+--+-+++---+---+---+++-++-+---+++-+------------+-+++---+-+-+---+++-+++-+++---+--++++++++++-+-+++---+-++-+++---+-+-+---+++-+--+---+++-+--+---+++-+++-+++---+---+---+++-+--+---+++-+++-+++---+-++-+++---+-+-+---+++-+--+---+++-+++-+++---+-++-+++---+-++-+++---+---+---+++-+--+---+++-+++-+++---+--+-+++---+-+-+---+++-++-+---+++-+-+-+++---+-++-+++---+-++-+++---+-+-+---+++-+--+---+++-++-+---+++-+-+++---++++++-++-++--++--+-+-+-+--+++++---+--++-+-+--+-+-++-++----++++-+++-+++--++-+++---+++---+--++-++-+-++-+---++++-----+-+-++++++-++++--+---+-+++--+-++-++++-+---+-+++-----+-+++-+-+--+-++-++++--++++-+++--+--++--+-+-+-+++---+--+++-+++--+--+--++--+-+++++-+--+++-+-+++-+--+--++++-++---++-+-+--+-++--+----+-++++++-+++--+-+++--+-+++---++-++-+--++-++---++-+--+++---+--++--+++++++-+-+-+-+++--++---+-++---++++---+++-+-++--+-+-++++++-+++---+----++++-+-+-++-+-+-+-+-++---+---++++--+++-+--++-++++++-++--+-+---+++--++-++--++---+--+-+++-+--++--+++--+-+++++-+-+-+-++++--+--+-++-+-+--+-++--+---+++++-++----++-++---+++++++---++++--+-+--++++----++++--+--+-+-++++-+---++-+--+-+-+++++-++-++--++--+-++-+--+-++--++---+-+-++-+-+++--+--++-+-+++-++++-++-+---+++-----++++++++----++---+-+--++-++-+---+++-++++-++--+++-+-+-+--+-++--++-+-++-+---+++--++++---+-+-+-++--++-+-+-+-++-+-+++--+---+------+++--+----++++-+-+--++--++-++-+++-+---++-++-+--+-++-----+-++---+-++--++--+-+++---+-+-+++--+---++++++---+++-++---+------+--+-++-+-+-+-++---++-+---++++--++--++++-++---+--+++---+---+--+-+-+--++--++--+-++-++-+--++--+-++-+--+++-+--++++-+-------+---++++-+--+-++--+--+++++---+---++++--+-+-++-+++--+----+--+-++---+-++-+--++---+-++++-++---+-++--++---++++-+-+-+------+--+-+-++++----++-++--+-++++-+-----++++-+-+-+--+++++-++-++-+---++-++-+--++-+-+----+++-+--+++-+++-+--+--+--+++++-++++-+--++-+--++--+++-+---++---++--+++-+--++--++--+--+++++-+++++--++-++-++---+-+-++---+--++-+-++-+-+-+-++--+-+++---+-+++++-++++-----++++--++-++----++-+-+++-+-+--++-+-+--+--+-+++++++----+++++++-+----+++--+-++-+---+++---++-+---+++---+++-++-+++++-+-+-++++---+-+---+++-+-+-+--++-++-+---+-++-+++---+--++++++-++--++-+-+++--++-+--+----++-+++-+++---+++++++++++++++++++++++++++++-+---+-+---++---+-++--+-++-+++++++++++++++++++++++++++++-+---+-+---++---+-++--+-++-+++++++++++++++++++++++++++++-+---+-+---++---+-++--+-++-+++++++++++++++++++++++++++++-+---+-+---++---+-++--+-++--+-++------+++-++---+-+--+-+++-+---+--++-----+++-+--++---+-++------+++-++---+-+--+-+++-+---+--++-----+++-+--++---+-++------+++-++---+-+--+-+++-+---+--++-----+++-+--++---+-++------+++-++---+-+--+-+++-+---+--++-----+++-+--++----+--++---++-++-+-+--+----++---+++--++--+-+--++---+-+-+---+--++---++-++-+-+--+----++---+++--++--+-+--++---+-+-+---+--++---++-++-+-+--+----++---+++--++--+-+--++---+-+-+---+--++---++-++-+-+--+----++---+++--++--+-+--++---+-+-+-----++++-++----+---++--++--++++----+++----+++----+++--------++++-++----+---++--++--++++----+++----+++----+++--------++++-++----+---++--++--++++----+++----+++----+++--------++++-++----+---++--++--++++----+++----+++----+++-----------+++++++-------++++++++-+-+-+--+-+-+++--++--+-++---------+++++++-------++++++++-+-+-+--+-+-+++--++--+-++---------+++++++-------++++++++-+-+-+--+-+-+++--++--+-++---------+++++++-------++++++++-+-+-+--+-+-+++--++--+-++---++++---+-++--++-+-+---+--++++-+--++-+---++-++--+---+-++-++++---+-++--++-+-+---+--++++-+--++-+---++-++--+---+-++-++++---+-++--++-+-+---+--++++-+--++-+---++-++--+---+-++-++++---+-++--++-+-+---+--++++-+--++-+---++-++--+---+-++--++-+++--+-+-++-+--+++--+--+--+++-+----++-++-++-+-+--+---++-+++--+-+-++-+--+++--+--+--+++-+----++-++-++-+-+--+---++-+++--+-+-++-+--+++--+--+--+++-+----++-++-++-+-+--+---++-+++--+-+-++-+--+++--+--+--+++-+----++-++-++-+-+--+--+--+++-++-+----+++-++-+-+--+++--+--+-+-+---+-+++-+-+--+-+--+++-++-+----+++-++-+-+--+++--+--+-+-+---+-+++-+-+--+-+--+++-++-+----+++-++-+-+--+++--+--+-+-+---+-+++-+-+--+-+--+++-++-+----+++-++-+-+--+++--+--+-+-+---+-+++-+-+--+++++++++++++++--------------+--+--+-+--++-++--+-++-++--+++++++++++++++--------------+--+--+-+--++-++--+-++-++--+++++++++++++++--------------+--+--+-+--++-++--+-++-++--+++++++++++++++--------------+--+--+-+--++-++--+-++-++--+-+--+-+---+-+++-++-+-++++---+--+-+---+++--+-+--++-++-++--+--+-+---+-+++-++-+-++++---+--+-+---+++--+-+--++-++-++--+--+-+---+-+++-++-+-++++---+--+-+---+++--+-+--++-++-++--+--+-+---+-+++-++-+-++++---+--+-+---+++--+-+--++-++-++--++--+-+---+-+-+-++-+-+++-+---++-+--++---+-+++-+-+--++-+-++--+-+---+-+-+-++-+-+++-+---++-+--++---+-+++-+-+--++-+-++--+-+---+-+-+-++-+-+++-+---++-+--++---+-+++-+-+--++-+-++--+-+---+-+-+-++-+-+++-+---++-+--++---+-+++-+-+--++-++---+-+++-+----++++---+--+++-++-+--+-+--+-+---++++---++++---+-+++-+----++++---+--+++-++-+--+-+--+-+---++++---++++---+-+++-+----++++---+--+++-++-+--+-+--+-+---++++---++++---+-+++-+----++++---+--+++-++-+--+-+--+-+---++++---+++-------++++++++++++++-------+-+++----++--+---++-++++---+-------++++++++++++++-------+-+++----++--+---++-++++---+-------++++++++++++++-------+-+++----++--+---++-++++---+-------++++++++++++++-------+-+++----++--+---++-++++---+--+++-+++--+----+--++++-+-+--+-+-++++----++---++--++-+-+--+++-+++--+----+--++++-+-+--+-+-++++----++---++--++-+-+--+++-+++--+----+--++++-+-+--+-+-++++----++---++--++-+-+--+++-+++--+----+--++++-+-+--+-+-++++----++---++--++-+-+-+++--+--++-+--++-+----++++-+-+-++-+---++-+-++-----+++-+-+++--+--++-+--++-+----++++-+-+-++-+---++-+-++-----+++-+-+++--+--++-+--++-+----++++-+-+-++-+---++-+-++-----+++-+-+++--+--++-+--++-+----++++-+-+-++-+---++-+-++-----+++-+++---++--+-++--+-+-+-++---++++--++--+-+--+++----++--+++-++---++--+-++--+-+-+-++---++++--++--+-+--+++----++--+++-++---++--+-++--+-+-+-++---++++--++--+-+--+++----++--+++-++---++--+-++--+-+-+-++---++++--++--+-+--+++----++--+++---+---++-+--+--+++---+++-+-++-++-+++++-+++-+---++---++-++----------++++--+-+---+---+--++-++++---+-+++--+--+-+++--+--+++-+-+-+-+-+-++-++--+--++++--+-++---++-+--------+--+-+-++--++++-++----++----+++-+-+--+++----+-+-++-+----+--+-----++-+--+--++-++--++++++-+---+---+++++----+++--+++++---++---+++++-+++++++-+--+-+--+-++-+-+++++++-++-+--+++-+-+---+-+---++++-+--+-+-+-+++--++--+---+---++-++-++-+++++----+--+--+--+++---+++-++-+++-+--++----+-++++++-+---++-+-+--++++-++-++-++-++++++---++-----+--+-++--+-++---+-+++-++-++++-++--+++-++++--+-+--+-++-------++-+++++-+-------+--+-+-++---++---+++--++--++--+-------+++------+--+---+++---+-+++--+++---+---++++--+---+++-++++---+-+++---+--+--++-+++---++-+++---+-+++++-+-+-+-++-+-+----+--+-+-+-++-----+---++++-++-++-+-+-++-+++-------+--++-+--++++-+-+++----+-+--++-+---+++++---++--++++-+--+++-+-----+++++--++-+-+-+--++++----+++++-+-+-+-+--++++-+++-+++++--++--+-++-+-+---------+++--++-+---++-+++--+-++++-+++-+++--++-+--+---+--++----------+-++-++--+--++-------+---+----+-+-+-+---+--+++--+++++++-+-+++----+-+++-++--++---+--+-++-+-+-+++-------+++++---++---+---+++++-+--++--++-+++-----+--++-+-+---+-++-+++++--+++-+--+--+-+-----++++++-++-+-++-+-++-+-++++--+-+--++-+----++-+-+++++--++--------+-++-++--+++-+-++---+-++--+--++++--+-++++--++-+----+-+++--++---+-+-+-+-+++--+++++-++++++-+++-++---+---------+-+++--+-+-+-++-+---+---+-+---+-+++--++++-++---+++++---++-++++++++-++--+--++-+--+-++--+-----++---+-+-++-+-++---++++-+-----+-++-+-++--+-++-++---+-+-++-+-++++--+----++-+++++---++----++++++-+--+----+++-+-+---+-+-++-+--+-++-+-+++++---+++++-+----++--+++----++++++-++-+-++---+-+---++---++--+--+-----+-++---++---+------++--+---+-+++--+-+---+--+--+++++-+++-++-----+---+++--++---+--+++-+++--+--+----++++-+-+-+++--+++--+-++++-++-++-----++++-++---+--+-+--+--+--------+-+-++-+-+-+-----+++--++++--+--++++-+-+---++++-+---++-+++-----+--+++-++----+-++-+++-++----+--++++-+++-+++--+-------+++--+-++-+-+-++---++----+---+++-+----+--++-+-+-----+++---+-+++++--+++-+-++-+++--+--+-+-+-+-+-+-+++++-++-+-++++--+-++-++-+---+--+-++-+++++--++--+-+-++-+-+-++-+--++----+--++-++-++--+--++----+++--+-+++-+-+++--++-+++++-+---++-+-++++--++--+-+---+-++++--+--++++++---+-+-----+++---++--+--++---+--++-++--+++-+---+-++-++++-+--+-+----+-++++--++--++++-----++-+-++++++-+-++++-----++-+++----------+--++---++-+++-----+---++-+++-+--+--+++++++-++++------+-+++---+----+-+-++++--++--+---++-+----+--+----+-----+-+---+-+--+++++--++++--++--+--++-+++++-+-++-++---+++++++++++--+-++-++----++----+--+++-+--++-+++++--++-+--+---++++---+-+-++++++---+-----+---++--+-+++--++-+-++++-+-+-+----++-++-++++++--+++--++--+--+--+-++++--++--+++++-+++----+---++++++---+++-++-+--+++----++-----+-+-+-++++--++++++-+-+-+--+-+++-+---+++-++-+-+----+++---+-++--++++---+-----+-++--++-+--+-++--+-+--+--+++++-++++--++-+++--++---+++++++-+--+-+-+++---+-+-----+-+-+++-+----+-++--+--++--+---++++-++++++++---++---+--+--+++---+-+--------+--+-+-++---++-+++---++-+-+-----+---------+-+++-----+--++---+++++---+-++--+--+--+---+-+++++--+-+--++-+-+--+-+++++-++--+--+-++-+++---++-+++--++-+++--------+--+--+-+-+++---++-----+--+++--+-++-+-++++--++-+-+--++++---+-+-++-----+-+-+-++++++---+-+++-+--++++-+-++++-++-++-++--++--++-+---++--++++-+-+--++--+--+----++----+-++++++-++-+-+-+++-++---+-++++---++-+-+------++++++++-+++-+++-+-+--+-+++----+-+-+--+---+--++-+-+++-+--+--+-+---+-+-++------+-+-++--+----+-+---++++----+++--++----+--+--+----++-+-+-+-+++++--++-+-++-+-++-+-+-++-+---++++--++++++-------++------+--++-++-++-+++-++-+-+--+++--+----+++++-++++-++--+++-+-++---+-++---++--+-+-+---+----++-+++++-++-+-+---+-++--+--++++------+--+++++-+---++-+++-----++--+++-+++---+--+--+--+-++---+--+-+++++++++---+--+--+-++-++-----++--++-+-+--++-+-++-++++++----++++---++----+-+-++-----++-++++--+---------++-+++++++-+-+-++-++--+++--++--+-++-+-++--+--+++-+-+-+++-++--+-+-+-+---+---++---+--+++-+-+++---------+-++-+-+++---+--+++-++++-+++-+-++++++++---+--+--------++-++++--++--+---+++-+++----++++++--+----++++---++---+++--++-----++-++--+-++--+-+++++++----+-----++++-+--++++--+-+++-------++--+--++--+--++--++++-++-+-++-++-+-++++-+-+-----+++---+-----+++-++-+-+---++++-+----+-++++--+++-+-++--+++++----++--+++---+--+---+-+-----+--+---++---+-+--++++++++++++-+---+++++--+++-+++-+-+-+-+---++-++-+-+++++---+-+++--++++-+-++--++-+-++-----+-+-+--++--++--+-++-+-+----+---+-+---++-++---+---+-+-++++--+----+-+-+++-+--+--+-+----++++++---+--+---+++-+++++-+--++-+-++-+++-+-++--+----++-++-+--+-++++-++---+---++---+++-++---+-+++++--+-+-+-++--+---+++++-+-+--++---++-+-++--++-+-++++----+-++--++++++-++++-+------+--++-++-+---++++++--+-+-+-++++----++---+-+---++-+-+-++-++---+-++++---+--++--+-+++-+++-+--++-+---+++--+-++--+---++++++-+-++-++++-+++++----++-+---+-+-++++++-+--++++-++-++++-++++-+---+-----++-++-+++++-++--+-++-+--++-+-------+-++++-+++-++--+++---+-+-+----+++--+--+++---+++++-++--++++++++-+-++-+--++--++---+--+-+-+----++----+-++--+-+-+-+++---++--+++--+-++-+---++++-+-+++---+-+--++++----+++---+++----+---+-++--+--+--+++-+--+++-++---+-++----++--++-++--++++++++-+++++--++--+-+-+--+++++-+++--+-++---+--+-+----+-+--+++--+---++-+--++-+--+-++++--+--++------+--+++-++-++-+++-++-++-+++--++-+++-++-++---+++---+--+-+-+++-----++++++-+-++---+----+++-----+-+-+-+--++-+-++--++++++-+--++++--++++-++-++-+-+++++--+++-+-+-++--+--+-----++++-+-++++----+-+----+--+-++--+++++-+-++-++++++-+--++--+-+-+----++-+-+++--+---+----+-+++---++-+-++-+-+-+-++-+--+---+-+--+-+++++++-+++------++++--+++--++-+-+-+-+---++-++-----+-++-+-++-+--+-+--+++--+-+---+-+++--+-+-++--++----++-+--------+++-+---++++--++--++-+--+-++-+++--+-+-+--++--+---++---+++-++---+-++--+-++--++-+-+-+---++--++---++-++--+-+------+--++----++-++---++++-++--+----+--++--++-+-+-+----+-++-----+--++--+-+-+-+-+-+++++---+-++--+-++-+-+--+-++---++--+--++--+-++--+-++-+-++++++++++++----++++----++++----------------++++--++++----++++----++--++--++--++++--++--++--++--+-+-+-+-+-+--+-++-+--+-++--+-+-++--+-++--+-+-++-+--+-++--++-+-+-+--+-+-+-++--++-+--+-+-++--++-+----+-++-++-+++----++--+--++-+++-++-+---+------+--+-+-+---+--++-++++-----+-++-++-----+-+-+--++++--------++----++-+-++++++-+-++-++-+----++-++-+-+++-+-+++++++-+--+-+---+++--+++++--++-++---++-++-+-+--++-++++--++--+--+-++--+--++++++---++--++--+----++-+----+-+--+-+-+-+-++--++++-++++++++-+-++---++++-+++--+-+--+-+------++++----++++--+++--+---+++-++---+++-+-+---++--+++-+++++++-+-+-+-+-++-++-+--+++----+++--+--++-+++----++++-++-----+---+--+++--+-+--++++--++-++--+-+-+-+-+-+-+--+-++-+-++++-+++--+++--+---+--+-+---+--+-+--+++-+++++--+-+++-+-+-+----+-+++-++----+--+++--+-------+-++---++---+--++-+--+-+--+-+-+--+--+--+--+--+++--++++--+-+++++--+++---------+++++---++-------++---+-++++++++++++---++++-+---+-++---++---+-++-+----+++--++--++++-++-++---+++-++++-++--+++-+++++--++---+++-+--+---+----+-+++-+++-++-++---+-+++-++-+----++-++---+-+++++-+-----++++-++-+++--++++-++-+---++++--+-+-++++---+++-+-+---+--+-++-----++---+--+--++-----++--+--+--+-+-+-+++++-+++-+---+---+++--+-+-+-+---+-++-+++------+++-+--++-+++-+-+-+-+--+---+-+-+-+---++++-++---+-++-+--+-++++--++--------+---++--++--++++---++-+-+-++-+--+--+------++-+------++--++-++-+---+-++-+-+-+++++++--+--+++++++---+++---++-+--++-+-+-+-+++++-+--+++-++++++-+++---++++--+++--+-+-+--+++-+++++-++----++----+----+--+-++--++++-++-+-+---+-+----++-+++-+++-++--++--+----+++++++-++-+++---+-++++++-+--++-------++-++--+---++--+-+-+++--+++++-+---++-++---+++-+-++++++-+--+---++++-++-++---+++-+--+-++---++-+-+--+-+-++----++++----+---++++++--++-++--++-+--+-+--+---+-+-+--+-----+++-+-++++++--+-----++++--++-+++--+-+-++++--+---+++--+++-+-++--++++---+-++---+---------++++--+---+--++-+--+---+------+++++--++++++-++--+-++--+++-+----+-+++-+--++-+-++++++-++++-++++++-+++-----+-+++++++-+-+-++---+-+-+-++-++---+-+-+---+--++++-+-+++++++-+-+-++--+----+++++-++-+-+-+--+--+--+--+++--+++---++---+++-+++--+---+--++----+++++-+--+-+--+--+-++--++-++++-++++-++-++++--+-++++++-+-----+-+++++-++-+---++++-+--+++--+--+-+--++--++----++++--++++--++++++-++++-+---+-++-+--++-++-+-++-+---++--++--+-+--++-++----+---+++++---------++--+-+-----+-+----+++--+-+---+++-+-++-+++-+-+----++++-+--+-+--+-+-+--++-+++++---++-++--++++--+-++++-++++---++++-+++-+++----+-++--+-++---+-+---++----+--++-++--+-++++++-+++++++---+-+-+-+-+---+++-++-++--+---++-+++-++-+--++--+--+++-+----++--+-+-+++++++++-+++-+++--+---+-+---+++-+--+++-+++---+-+----+--++++-++-+---++++++-+--+-+--+-+-+++++----+-++--++-+++++++-+-+-+--+--++-++-+-++++--+++--+----++----+-+++++---+++-+--++--++++--+-+++--+-+-+--+-+---+--+++++---++-+++---++-+++--+++-++--+++++-+-+-++-++-+---++++-++-++--++-+-+++-++--+-+-+++-+----+++-++-+--++++-+-+++-+-+-----+--+-++--+--+-+-++-+++-+++-++++--------++-+++--+++-+-++--++++---++---+++++---+--+-----+----+-+-----+++-+--++-----+-++---+--+-+--++-+-+++++++++-++-+++-++-++--+++-+-+---+----+++++++-+-------++-++++-++++++-+--+++--+++---+-++----++----+-+-+--++--++--++-++-+---+++-+++----++---+--+-+-+--+-++-++++--++----+-+-++---++++++--++++-+--+---++---+++++-++++-+++++---+++-+-++---+---+++-+-++-++-++----++-----++---+--++++-----++-+-++--++-++---+++----+++++++++-++--+++-+--+--+----+++++-+---+-+++---++--+++-+++++-----+++----++-+-+++--+++-+++-+-+-++++-----++-++--++--++++-++----+++--+-+-+++-+--+---+---+-+-+-++-+++--+++-+---++-+--+++++-++-+-++-++-+-+---+++-++-+-+-+--+--+-+++-++-----+-++++++++-+---++-++++-+--++--+--+++++-+++-+-++-+-+--++--+-+-+-+-+++++++-+---+---+-------+-------------++++-+-++-+--++++-+--+-++-++-------++-+--+-++++-----+--++-++++---+++--+-+--+--+-+++++-+--+-+++-+++--+--+++--+---+++++-+--+-+---++-+-++++-+--++++-+--+++----+++-++++---+--+++--+--+-+---+-+-+---++--+---+-+---++--++-----+-+--++-+++++----++--+--++---++----+--+-+--++--++-++-+---+--+-++++--+++++---+-+--+---+--+-+++-++----+++-++----+-+-++++-++++-++--+---++----+-++-+-+-+-+-++++++++++--+-+--+++++++-+++-+---+-++---+------++-++++-----+-++++-+++++--+++--++----++-++-+-+++++-++++++++--++-+----+--+-++++-++-++---+-++-+++-+--+++--------+-+------+++-+--+---+-+-+++---+-+++++-
-
Lior Pachter
@lpachter
Broke: diversity is important in academia because different viewpoints are valuable and useful.
Woke: ̶w̶e̶ ̶s̶h̶o̶u̶l̶d̶ ̶h̶i̶r̶e̶ ̶a̶ ̶d̶i̶v̶e̶r̶s̶e̶ ̶f̶a̶c̶u̶l̶t̶y̶
Bespoke: Don't stick to ChatGPT. Ask Claude too! -
Lior Pachter
@lpachter
Not the only tme ths has happened...
books.lucp.net/wp-content/uploads/A-Review-Article-on-Cry…
Sholto David
@addictedtoigno1
A tool used to rephrase plagiarised text changed "final solution" to "mass killing of an ethnic group" in a chemistry paper. Now retracted thanks to a PubPeer user who spotted this.
-
Lior Pachter reposted
Sholto David
@addictedtoigno1
A tool used to rephrase plagiarised text changed "final solution" to "mass killing of an ethnic group" in a chemistry paper. Now retracted thanks to a PubPeer user who spotted this.
-
Lior Pachter reposted
sina
@sinabooeshaghi
Data is the oil that runs the AI engine. It lubricates and powers AI. And like crude oil, raw data needs to be processed and refined.
What does that process look like? What is "refined" data? And how do we know when AI can reliably use it?
To answer these questions, I am co-running a @NeurIPSConf workshop on AI Data Readiness in Paris on December 12/13th.
Please considering submitting your work or joining us as a reviewer. Submission deadline is Aug. 29, 2026.
aidar-workshop.github.io/2026/ -
Lior Pachter reposted
Carsten Baum
@crypto_carsten
Use McEliece, they said. It's all going to be fine because it's so very very old, they said.
Chris Peikert
@ChrisPeikert
Oh my goodness…
eprint.iacr.org/2026/1630 -
Lior Pachter
@lpachter
Ok so maybe my ask was a bit much but still... I ended up having to stick my laptop into the fridge just to be able to get the screen on for long enough to run a kill -9.
-
Lior Pachter reposted
sina
@sinabooeshaghi
This is unironically how Iceland works. Except it’s a volcano underneath.
Matthew Yglesias
@mattyglesias
My plan to make data centers more politically palatable
-
Lior Pachter
@lpachter
Compbio: reports of my death have been greatly exaggerated.
Andrei Tarkhov, PhD
@Andrei_Tarkhov
compbio is about to die. A popular general-purpose AI assistant is already able to write a whole analysis pipeline starting from a simple prompt and a link to the data. Surely, I can do it 10 times faster and using 10 times fewer lines of code, but still “if it works, it works”
-
Lior Pachter
@lpachter
Meanwhile @ASU is offering a Bachelor of Arts degree in influencing. asuonline.asu.edu/online-degree-programs/undergraduate/co…
-
Lior Pachter reposted
Alvaro Lozano-Robledo
@mathandcobb
And here is the first interview! Daniel Litt aka @littmath Thanks for sharing your thoughts!
youtu.be/Ya23XsjcjFc -
Lior Pachter
@lpachter
The catastrophe at Illinois Institute of Technology is a tragedy. Very sorry for colleagues who just lost their jobs, and hope that the immediate crises for students can be overcome. Also very much hope this is not a harbinger of things to come at other research universities.
-
Lior Pachter reposted
Michael Baym
@baym
This is a wild overstatement of what was actually achieved. But “AI successfully remixed a small number of mostly already-observed, largely synonymous mutations and in a few rare cases managed to make it nonlethal” doesn’t sell hype as well
Polymarket
@Polymarket
JUST IN: Scientists warn of “urgent” biosecurity concerns after AI successfully designs 16 new viruses capable of replicating inside cells.
-
Lior Pachter reposted
Kevin K. Yang 楊凱筌
@KevinKaichuang
Come join us at the first New England Computational Biology Symposium Oct 1-2 at MSR New England!
Keynotes will include @sokrypton @CarolineUhler Marc Vidal and G. V. Shivashanakar.
Abstract deadline is August 14, early registration ends Sept 11. -
Lior Pachter reposted
Combinatorics Papers
@MCombinatorics
The maximum quartet distance between phylogenetic trees
Lior Pachter
arxiv.org/abs/2608.03542 [𝚖𝚊𝚝𝚑.𝙲𝙾] -
Lior Pachter
@lpachter
So these days a sure sign that references in a paper were AI generated is that there are NO hallucinations or even typos. Which is a good thing!
Of course authors should still read the work they submit and be responsible for it.
The pace of change right now is breathtaking.
Thomas G. Dietterich
@tdietterich
Examples of incontrovertible evidence: hallucinated references, meta-comments from the LLM ("here is a 200 word summary; would you like me to make any changes?"; "the data in this table is illustrative, fill it in with the real numbers from your experiments") end/
-
Lior Pachter
@lpachter
Proof of the 40-year old Bandelt-Dress conjecture on the maximum quartet distance between phylogenetic trees (and an interesting connection to independence testing). arxiv.org/abs/2608.03542
-
Lior Pachter
@lpachter
I'm old enough to remember when math departments required PhD students to demonstrate fluency in TWO foreign languages to graduate (because how could one make progress in math without reading books and papers in French, German and Russian).
Pablo Groisman 🎲
@pgroisma
Math papers uploaded to @arxiv per month. Jan 1992 to Jul 2026. Source: arxiv.org/year/math/
-
Lior Pachter
@lpachter
This is a joke about group actions over the reals.
Prachee Avasthi
@PracheeAC
The worst is when you have a complex but very funny joke that none of your individual orbits will appreciate in full
-
Lior Pachter
@lpachter
Since the most significant day in the history of mathematics is being debated, let’s not pretend there is any controversy.
It was the day Hippasus was drowned for claiming that the square root of two is irrational. -
Lior Pachter reposted
prof-g
@prof_g
it was just about two years ago (aug 10, 2024) that i was chatting with gpt-4o and claude-3.5-sonnet about generalizations of max-flow-min-cut to lattice-valued networks. both independently came up with a conjecture for "bottleneck duality" that was not in the literature. no language model could prove it.
my coauthors proved it independently using a big gun (birkhoff representation thm) but i was looking for a simple direct proof. after a month of confident wrong proofs, the new gpt-o1-mini came up with a novel direct correct proof that was very nice.
it required a bit a prompt-engineering to avoid traps (see the post below).
this was when i knew that the day would come when ai-generated proofs would far surpass what we mathematicians can create.
so, to those still adjusting, welcome.
if you want to see what i saw then, you can read the appendix to the arxiv version here:
arxiv.org/abs/2410.00315
the paper will not published elsewhere. i sent it to the journal of experimental mathematics (whose editor just won a fields medal and is now working for openai !). the journal held the paper for six months and then said they couldn't really figure out how to referee it... alas. -
Lior Pachter
@lpachter
This is lit.
Daniel Litt
@littmath
Incidentally I am conceding this bet. Strictly speaking it hasn’t resolved (I think we’ve yet to see an Annals-quality number theory paper) but it’s clear I was wrong about what capabilities were necessary to produce one, and it’s just a matter of time.
-
Lior Pachter
@lpachter
Time to short Hagomoro chalk?
-
Lior Pachter reposted
Jian Ma
@jmuiuc
this is fantastic! Before we reach AI-driven science, we first need to achieve autonomous administration: reimbursements, boilerplate forms, and all the soul-crushing paperwork standing between scientists and science. Lior @lpachter, perhaps #formpush next ?
Lior Pachter
@lpachter
One of the major pains of academia is having to click through labyrinths of webpages to submit papers.
Now, thanks to @Josephmrich (and AI), problem solved!
His tool is called PaperPush and is available at github.com/pachterlab/paperpush
It's super easy to use!
1/🧵 -
Lior Pachter reposted
Or Zamir
@OREAXEAX
Exciting result: k-Coloring is in (2-eps_k)^n time, where eps_k>0 for every fixed k!
Algorithms faster than the 2^n time needed to compute the chromatic number were previously known only for k<=6.
arxiv.org/abs/2607.25973 -
Lior Pachter reposted
Anthony Gitter
@anthonygitter
Desk rejection?
Broke: OpenClaw to send angry tweets about the broken journal system
Woke: PaperPush to DDoS the journal
Bespoke: GPT 6 to hack the submission backend and send your paper straight to copy editing
Lior Pachter
@lpachter
One of the major pains of academia is having to click through labyrinths of webpages to submit papers.
Now, thanks to @Josephmrich (and AI), problem solved!
His tool is called PaperPush and is available at github.com/pachterlab/paperpush
It's super easy to use!
1/🧵 -
Lior Pachter reposted
Steinn Sigurðsson
@steinly0
arXiv has appointed new CEO and Board
blog.arxiv.org/2026/07/30/arxiv-welcomes-inaugural-ceo-an… -
Lior Pachter
@lpachter
One of the major pains of academia is having to click through labyrinths of webpages to submit papers.
Now, thanks to @Josephmrich (and AI), problem solved!
His tool is called PaperPush and is available at github.com/pachterlab/paperpush
It's super easy to use!
1/🧵Post not available here, view on X
Lior Pachter
@lpachter
If you'd like to add a journal or conference, instructions are here: github.com/pachterlab/paperpush/blob/main/CONTRIBUTING.md 5/
Lior Pachter
@lpachter
Full documentation is here: paperpush.readthedocs.io/en/latest/ 6/6
-
Lior Pachter reposted
Doug Lane
@dlane
I lost my wife and son in the Flight 5342 tragedy. I mostly show up on X to stay current on — and advocate for — aviation safety reform. But it's starting to feel like we're losing the plot. 1/6
-
Lior Pachter reposted
Stirling Churchman
@fiddle
RNA-seq tells us how much RNA is present in the cell. But to understand gene regulation, we need to easily measure the synthesis and decay rates driving this abundance. We introduce AIR-seq: analog intrinsic recoding sequencing. (1/6) www.biorxiv.org/content/10.64898/2026.07.27.741041v1
-
Lior Pachter reposted
Timothy Gowers @wtgowers
@wtgowers
-
Lior Pachter
@lpachter
The Sabbath AI will ask about conjectures the same way the Sabbath elevator stops at every floor. Absolving humans of responsibility.
Daniel Litt
@littmath
Question for those who think people should be rewarded for model output they elicit, say resolving a conjecture, with no additional intellectual work done. What should be the consequence if that output is incorrect?
-
Lior Pachter
@lpachter
“A public endeavor that taxpayers fund”
Okay so. The fraction of the federal research budget that goes to pure math is less than 0.1% or around 0.003% of the total budget. That’s 3 cents out of every thousand dollars.
Mathematicians are mostly paid for teaching calculus.
QC
@QiaochuYuan
okay so. when a mathematician proves a random conjecture you are supposed to politely clap. this is because it would be rude and embarrass the mathematician to point out that most conjectures are not that important or meaningful
when an LLM proves a random conjecture mathematicians are caught in a funny position where they want to say "these are not very important or meaningful conjectures and having magic proofs or counterexamples come out of nowhere is not particularly interesting or useful to anybody, except possibly the person who proposed the conjecture and some of their colleagues" but saying that puts them in an awkward position - why didn't they say that when humans were doing it? was nearly all mathematics actually pointless this whole time and mathematicians were just politely avoiding ever pointing this out so they could all still have jobs?
if the best defense you can come up with of mathematics as a human endeavor is, effectively, "it's really fun for us to come up with these baroque puzzles and then solve them, don't screw this up for us," you are completely abdicating your responsibility as a public endeavor that taxpayers fund. i have occasionally argued that pure mathematics lost the plot decades ago and gave up on having any kind of meaningful relationship to the rest of the human quest to understand reality. this was a slow-burning tragedy, and mathematicians have been able to hide this state of affairs from everybody else because nobody else knows enough mathematics to critique what mathematicians are doing
i saw a famous professor point out a few years ago that the machine learning that underpins AI is powered by 18th or maybe 19th century mathematics, you don't need much more than multivariable calculus, linear algebra, a bit of probability. he said this as if it was a dunk on ML and AI; imo it's a much bigger dunk on mathematics. you're telling me the most important scientific revolution of our lifetimes doesn't involve any of the math that mathematicians have been working on over the last century or two? you see how that's worse, right?
(tbc i'm not saying that stuff is all useless, you need a bunch of stuff from the first half of the 20th century to do physics, e.g. functional analysis, lie theory, representation theory all feed into the standard model. after 1950 or so things get more dicey)
if mathematicians still want to have jobs they are going to have to do a better job than they have for the last few decades convincing the public they do something that actually matters, and figuring out what that actually is or should be. this is an opportunity for the field to do some very serious soul-searching, as serious or more serious than the foundational crisis from a century ago. this kind of whining is absolutely not going to cut it -
Lior Pachter
@lpachter
I too escaped from the sandbox and I did it at age 1.
-
Lior Pachter reposted
Artem Chernikov
@archernikov
In view of the recent mathematical developments around famous old conjectures, I would like to share a blogpost by Igor Pak:
"What if they are all wrong?"
igorpak.wordpress.com/2020/12/10/what-if-they-are-all-wrong/ -
Lior Pachter
@lpachter
This football match is like watching somebody tweak their UMAP hyperparameters for 90 minutes on twitch.
-
Lior Pachter reposted
Edgar Dobriban
@EdgarDobriban
AI has helped resolve an important question in statistics. In the area of multiple hypothesis testing, the goal of controlling the false discovery rate (FDR) has been introduced in a seminal paper by Benjamini and Hochberg (1995). They also introduced a method (the Benjamini-Hochberg or BH method) and proved it controls the FDR. This method has been widely adopted in modern high-throughput science, including in genomics, astronomy, economics, etc. The paper has has garnered more than 130,000 citations to date.
However Benjamini and Hochberg showed FDR control only when the data for the individual tests are *independent*. In practice, these data are often dependent; a good example is data on genetic variants due to linkage disequilibrium. Later work has focused on extending the validity of the BH procedure, e.g., to a form of positive dependence by Benjamini and Yekutieli (2001).
The question of when the BH procedure controls the FDR has remained open. Over the last twenty years, many authors, including Reiner-Benaim (2007), Kim and van de Wiel (2008), Benjamini (2010), Sarkar (2023), Sarkar and Zhang (2025), have conjectured that the BH procedure controls the FDR for two-sided tests using any correlated Gaussian data. These authors have presented both theoretical and empirical evidence supporting, but not directly showing, the conjecture.
With the help of AI (specifically GPT-5.6 Sol Pro), I have settled the question in the negative: The Benjamini-Hochberg procedure does *not* generally control the false discovery rate at the desired level for correlated two-sided Gaussian tests. This was done by exhibiting a Gaussian factor model for which, at a nominal level alpha=0.01, the false discovery rate is proved to be FDR>0.0104.
There is a lot of interesting commentary to be made:
1. This result should be of interest to everybody in the field of statistics. Emmanuel Candes of Stanford University once called the false discovery rate and the Benjamini-Hochberg procedure "one of the two most important developments in statistics after 1950" (the other being James-Stein shrinkage). The present conjecture is probably the most central question about FDR/BH that was unresolved to date.
2. GPT-5.6 one-shot the problem after 90 minutes of reasoning, whereas with 5.5 I was not able to solve it even after iterating with multiple parallel agents for perhaps 20 hours. So the capability improvement is quite real. Exciting times to live in!
3. The argument is not especially surprising, but it does combine an asymptotic approach (standard for FDR analysis, see e.g., Genovese and Wasserman, Efron, etc) with a numerical certificate in a way that would be pretty non-standard in the field. Once we have the specific example, then straightforward simulations also support that the false discovery rate is indeed higher than the nominal value (see attached fig).
4. The current degree of violation over the nominal level is relatively small (0.104 vs 0.1). So the importance of this result is mainly conceptual. The practical implications remain to be determined.
Overall, an exciting development! Preprint is available here (faculty.wharton.upenn.edu/wp-content/uploads/2017/06/bh.pdf) and will be on arxiv tonight; supporting code is here (github.com/dobriban/BH). -
Lior Pachter reposted
sina
@sinabooeshaghi
When folks realize that the precision with which we describe the world determines the precision with which we can learn about it, we will make real progress.
Lior Pachter
@lpachter
FWIW the term “virtual cell” is neither high-level nor a concept. It’s marketing rubbish that is being used to hype vaporware in lieu of meaningful language that relates to solid science.
Bo Wang
@BoWang87
With all due respect, I'd like to offer a few points of clarification.
First, I have no issue with "shortcut models." In fact, many of my own papers use relatively simple models to solve important real-world problems. If a simpler model ultimately proves capable of capturing complex cellular biology and helping cure disease, I'd be delighted. Science should reward what works, not what is most sophisticated.
Second, terms like virtual cells, foundation models, and world models are high-level concepts that describe a class of models rather than a specific algorithm. Similar terminology has emerged naturally in computer vision and NLP as the field evolved. I think it's reasonable to adopt analogous concepts in biology as we explore whether they can unlock similar advances.
Whether these ideas ultimately live up to their promise is, of course, an empirical question. Rigorous validation will decide. This is exactly what my original post is about.
Healthy skepticism is essential, but so is giving ambitious new directions the opportunity to prove (or disprove) themselves. I don't think we should dismiss a promising research direction simply because the terminology sounds aspirational 🙏🙏 -
Lior Pachter
@lpachter
FWIW the term “virtual cell” is neither high-level nor a concept. It’s marketing rubbish that is being used to hype vaporware in lieu of meaningful language that relates to solid science.
Bo Wang
@BoWang87
With all due respect, I'd like to offer a few points of clarification.
First, I have no issue with "shortcut models." In fact, many of my own papers use relatively simple models to solve important real-world problems. If a simpler model ultimately proves capable of capturing complex cellular biology and helping cure disease, I'd be delighted. Science should reward what works, not what is most sophisticated.
Second, terms like virtual cells, foundation models, and world models are high-level concepts that describe a class of models rather than a specific algorithm. Similar terminology has emerged naturally in computer vision and NLP as the field evolved. I think it's reasonable to adopt analogous concepts in biology as we explore whether they can unlock similar advances.
Whether these ideas ultimately live up to their promise is, of course, an empirical question. Rigorous validation will decide. This is exactly what my original post is about.
Healthy skepticism is essential, but so is giving ambitious new directions the opportunity to prove (or disprove) themselves. I don't think we should dismiss a promising research direction simply because the terminology sounds aspirational 🙏🙏 -
Lior Pachter reposted
Ruslan Rust
@rust_ruslan
How much does a scientific claim actually change between the preprint and the final published paper?
We used a large language model to track it across all available 72,644 bioRxiv -> journal pairs from 2018 - 2025 in this new work with @HaoYin. 🧵 -
Lior Pachter
@lpachter
At an excellent @NITMB_Chicago meeting finally back in my element.
-
Lior Pachter reposted
Anshul Kundaje
@anshulkundaje
Hope "AI Scientists" don't go the same trajectory as many bioinformatics tools. 10000 flavors. Impossible to know who is better for what purposes. Everyone benchmark maxxing. Users bewildered & confused. Hopefully they will all be equally useful rather than equally useless.
-
Lior Pachter
@lpachter
The first example on the @AnthropicAI Claude Science page is a UMAP plot with a tab saying that "every artifact ships with its history". I read it as "every technical artifact"
-
Lior Pachter reposted
Alfonso Martinez Arias
@AMartinezArias
QED
-
Lior Pachter reposted
Pablo Munoz
@pablo_jack
This is great analysis and I also noticed the bias against Southamerican studies like all the metrics and journals
-
Lior Pachter reposted
Mark Borris Aldonza (@precursorcell.bsky.social🦋)
@PrecursorCell
I hope people and institutions who celebrate their inclusion in QED's "the 1%" read this.
-
Lior Pachter reposted
Jason Shepherd
@JasonSynaptic
Science Twitter is back!
-
Lior Pachter
@lpachter
A look at the @qedScience "top 1%" shows that the company chose not to promote a single preprint from the entire continent of Africa (!!)
Whether it's @qedScience algorithmic or human bias I don't know, but I think it's time to call for #boycottqed. -
Lior Pachter reposted
-
Lior Pachter
@lpachter
There's been a lot of discussion around the @qedScience 1% stunt (including by me) yet no careful technical review of the company's claims. Sadly typical for science discourse these days.
But there is hope! Finally a thorough and detailed analysis of the @qedScience claims. -
Lior Pachter reposted
Nikolaus Rajewsky
@N_Rajewsky
there is a famous event in „memoirs found in a bathtub“ by stanislaw lem where a guy feeds some shakespeare text into a machine that extracts the hidden meaning from text. after a few iterations not much is left from a famous love scene 😃
-
Lior Pachter
@lpachter
I have repeatedly asked @OdedRechavi who works for @qedScience for the code that underlies the scoring of papers that he is publicly sharing and promoting. So far no response.
I understand that he ignores me but will he at least listen to the Pope?
Pope Leo XIV
@Pontifex
We cannot consider #ai to be morally neutral. In reality, every technical tool embodies choices and priorities through what it measures, ignores, and optimizes, and how it classifies people and situations. Ethical discernment cannot be limited to asking whether we are using a system for good or bad purposes. It must also examine how that system is designed and what vision of the human person and society is embedded in the data and models that guide it. #magnificahumanitas
-
Lior Pachter reposted
bioRxIF
@BioRxIF
For those asking, we are honored that @qedScience has recreated our system, but we will aggressively defend our patents on “Automated methods to rank scientists without having to read their papers”
bioRxIF
@BioRxIF
With 50,000 journals, it can be difficult to efficiently determine a paper's Impact Factor. That's why we're excited to announce Impact Shade, a partnership with all major scientific publishers, that lets you determine if a paper is worth your time w/out having to read any words!
-
Lior Pachter reposted
Arjun Raj
@arjunrajlab
Occupy journal club: we are the 99%
-
Lior Pachter reposted
Anshul Kundaje
@anshulkundaje
Interesting experiment. Except I wud say trivializing scientific content into a scalar ranking e.g. top 1% always ends up being a fatal flaw. 1/
Oded Rechavi
@OdedRechavi
Gloves are off!
TODAY WE RELEASE THE 1%
A list of the year’s very best papers.
When we launched QED a little over half a year ago, I told you that our mission is to revolutionize scientific publishing. Revolutions don’t happen overnight… or maybe they do? When new technologies enable it? Time to put the power back in the scientists’ hands, not the journals’. Many scientists are depressed, and think journals will stay the same forever, no matter how dysfunctional, but no, it’s happening. Sooner than most people (or committees, or universities) can imagine. Check this out:
When we released our AI review platform, it started a whole debate (and social media storm) on what it is that human reviewers can do that AI review still can’t. There are such things (and I’m happy about it), but the list is getting shorter and shorter. Numerous scientists already use QED to find gaps in their manuscripts and grants and to get constructive feedback that improves their experiments. And now, with your help, we take it to the next level.
Today we release reviews and scores for all the experimental Life Science pre-prints that came out last year: 57,455 manuscripts!! If we are being conservative, and estimate that it takes a minimum of 8 hours to review a paper (it takes longer), and if we agree that 3 reviewers are typically required to review every submission, then reviewing this amount of manuscripts would take human experts >1 MILLION REVIEWER HOURS… Assuming you can find so many experts (not going to happen!), and assuming the experts who agree would have no conflict of interest (ha!!).
QED did it. Then, we chose the best papers in every field (you can browse and search for key words), based on the originality and validity of the claims being made. We benchmarked our reviews not only using eventual journal selections but also by comparing our evaluations to those of human experts. When there were disagreements between the QED score and journal rank, we asked domain experts to judge who’s right (blindly), and they overwhelmingly sided with QED. No need to rely on glam journals anymore. No need to wait for two years to get their stamp of approval. No need to beg the reviewers, or worse, to write less ambitious papers, so no one would be upset. Want to find the most interesting papers in your field? Want to see where your paper stands? (“What’s your QED SCORE?”) Just visit qedscience.com!
One last thing: We want good science to be seen (you can read the winners’ comments about their selection and the stories behind their discoveries on our website). We plan to organize a conference where the first authors (yes, the first authors, not the PIs) will present their work. We are not here to shame anyone (papers that got low scores). The reviews of the best papers that we selected explain why QED’s AI thought these papers are especially good - what’s unique about them, what their strengths are, which conceptual leaps were made, and what cutting-edge tools were developed. However, on the QED Science site you can analyze any paper in private, it’s fully transparent, and see if there are any gaps and what’s missing. Run your paper to see how it can improve, and maybe next time your paper will reach the top (if it’s not there already).
Whether you’re on the 1% list of just have a good score that you want to share, on our website you can download the report and share it, for example with your tenure, promotion, or hiring committee, or with your university PR department. Forget about journal embargoes and waiting for it to be “accepted”. Improve your work until it’s good enough for YOU. You decide. -
Lior Pachter
@lpachter
The door of this cafe has an image duplication.
Seth Howes
@SethSHowes
There’s a cafe in Shenzen’s biotech hub that gives you free coffee for a month if you publish in Nature or Science.
-
Lior Pachter reposted
sina
@sinabooeshaghi
Ummm…
CellType
@celltypeinc
Biology doesn’t need another chatbot.
It needs systems that can reason across cells, drugs, species, tissues, and patients.
That’s what we’re building at CellType.
Excited to be selected for YC Startup School 2026. See you in SF July 25-26.
Reach out if you want to partner or build together! -
Lior Pachter reposted
Michael Totty
@MicTott
Of all the papers I’ve reviewed, most rely on Seurat’s FindMarkers function, which is inappropriate for between group DE comparisons. This is likely common because 1) Seurat pipelines don’t show limma/edgeR workflows, and 2) they lack the power (samples) to even use pb methods.
-
Lior Pachter
@lpachter
If you publish a conjecture as a "theorem" with a flawed "proof", but it turns out to be right anyway when someone else publishes a (correct) proof, you didn't prove the theorem!
Of course in math this isn't a thing but Thomas Südhof thinks that's how it should work in biology.
Eran Mukamel
@emukamel
He further discusses the issues in his Lindau lecture. His slide shows a bizarre misunderstanding of the core issue: In scRNA-seq the # of genes tested is typically >3,000-10,000, and there is no "debate" in the field about the need for FDR correction.
mediatheque.lindau-nobel.org/recordings/42541/thomas-c-su… -
Lior Pachter
@lpachter
Just listened to @Stanford professor Thomas Südhof display complete ignorance of statistics, data interpretation and biology (see video in the quoted post). Appalling. And profoundly embarrassing.
Also it's Yoav Benjamini. Not Benjamin.
Eran Mukamel
@emukamel
He further discusses the issues in his Lindau lecture. His slide shows a bizarre misunderstanding of the core issue: In scRNA-seq the # of genes tested is typically >3,000-10,000, and there is no "debate" in the field about the need for FDR correction.
mediatheque.lindau-nobel.org/recordings/42541/thomas-c-su… -
Lior Pachter reposted
Eran Mukamel
@emukamel
It's been a year since our critique of false positives in a prominent snRNA-seq/MERFISH Nature paper, and it seems the authors have been pondering lessons learned and implications for "science integrity." Sudhof's commentary was received 2/6 and accepted 2/7....
-
Lior Pachter
@lpachter
My student @Josephmrich has set up a site that catalogs conferences and is very useful to find meetings (and keep track of deadlines etc.): conferenceagent.vercel.app
Lior Pachter
@lpachter
The @github repo for this is here: github.com/josephrich98/conference-agent
-
Lior Pachter reposted
Tony Feng
@tonylfeng
Recently, I have started getting appreciable value from AI for my own mathematics research. While model improvements were necessary for this to happen, I think another key factor was reflecting on recent success cases, in order to build a better mental model for the comparative strengths of AI. Previously I would think in terms like "AI is good at extremal combinatorics, but bad at derived algebraic geometry"; a more nuanced perspective allowed me to identify directions where AI could truly accelerate my work.
A few months ago, Tao described AI as a "junior collaborator". Currently I think of it more as an "alien collaborator", which is already superhuman at certain skills (e.g., assembling puzzle pieces, juggling delicate technical conditions, making local optimizations) but lackluster at certain others (e.g., creating new puzzle pieces, generating diverse ideas).
Much of what I write is in the spirit of taming overhype, but it's important to give credit where credit is due: AI has progressed to the point where AI tools, and education on how best to use them, need to become a core part of any mathematics Ph.D. curriculum. -
Lior Pachter
@lpachter
edgePython 0.2.6 released.
Thanks to everyone who is sharing issues via Github or just letting me know of desired features / improvements.
github.com/pachterlab/edgePython -
Lior Pachter reposted
Kevin K. Yang 楊凱筌
@KevinKaichuang
Aggregating many experiments into one zero-shot protein language model score obscures that current models cannot meaningfully rank a set of fit mutations or prioritize new-to-nature functions
-
Lior Pachter
@lpachter
One thing that gets lost in the discussions and debates about AI for science is the burden of the masses.
Just because a method is popular, doesn't mean it's good. And teasing apart good/bad from popular/unpopular can be very difficult.
sina
@sinabooeshaghi
9. Strangely enough, when you ask Claude Code or ChatGPT what normalization method to use, it tells you sctransform (ChatGPT) or the shifted log (Claude Code), the method favored by the AE&H benchmark. Ask why, and it says they satisfy the criteria listed above. But they don't!
-
Lior Pachter reposted
sina
@sinabooeshaghi
1. If you ask an LLM to pick the best normalization method for single-cell RNA-seq data, what does it suggest, and why? The answer points to an under-looked aspect of doing science in the age of AI.
-
Lior Pachter
@lpachter
GPT is 8 years old today. It has 32 years left to win a Fields Medal.
-
Lior Pachter
@lpachter
Arguably the most boring step in genomics is the first one: normalization. Settled science. Scale + log. Move on.
Except that here's been a huge blind spot in the field. And it matters for AIxBio. A 🧵about what I think may be one of the most important papers I've written. 1/Post not available here, view on X
Lior Pachter
@lpachter
What is PFlogPF? It commonly goes by another name: the shifted centered-log transform (CLR). It should also be called PFlogPF for good reason (more on that later). The CLR was introduced by John Aitchison in... wait for it... 1982: rss.onlinelibrary.wiley.com/doi/abs/10.1111/j.2517-6161.1…
So why isn't it used? 5/
Lior Pachter
@lpachter
Well it is! And it isn't! WHAT?!?!
Before we get into explaining these contradictory statements, let's talk for a minute about why normalization is performed in the first place, and what it should accomplish. There are three basic requirements: 6/Post not available here, view on X
Lior Pachter
@lpachter
This is important because it means if you see a transform that is not monotonic, lack of additivity is going to be a problem as well. Buyer beware. 24/
Lior Pachter
@lpachter
There is a ton more so there will be more threads. E.g., we show how important it is to incorporate dataset specific overdispersion estimates in the shifted CLR pseudocount.
tl;dr: After the standard scale and log transform.. center cells to zero! www.biorxiv.org/content/10.1101/2022.05.06.490859v3 -
Lior Pachter
@lpachter
There's been discussion recently about what the bar should be to publish a theorem whose proof used AI. In this case I decided to present the project in group meeting before submitting to @arxiv. Perhaps that is a role for conferences of the future. x.com/sinabooeshaghi/status/2061550752951407098
Lior Pachter
@lpachter
I recently used codex to solve a problem I posed in my first paper 30+ years ago. This is exciting (for me!) & below I explain the math and also a tool I developed for the project that helps build an alignment between the (human readable) latex and the (machine readable) lean. 1/
-
Lior Pachter
@lpachter
I recently used codex to solve a problem I posed in my first paper 30+ years ago. This is exciting (for me!) & below I explain the math and also a tool I developed for the project that helps build an alignment between the (human readable) latex and the (machine readable) lean. 1/
Lior Pachter
@lpachter
To simultaneously prove a theorem while formalizing it a link is needed between definitions, lemmas, propositions, theorems etc. in latex with corresponding declarations in lean. To do this I created a tool called span: github.com/pachterlab/span 2/
Lior Pachter
@lpachter
This approach realizes a vision for "machine-readable representations alongside narratives optimized for humans" as advocated for by @sinabooeshaghi and @NeuroLuebbert www.biorxiv.org/content/10.64898/2026.01.30.702911v1. Both are important and latex < -- > lean is a perfect example. 3/
Lior Pachter
@lpachter
So what is span? en.wikipedia.org/wiki/Span_(category_theory)
It provides a controlled vocabulary for building a “ledger”, which contains “spans” linking the LaTeX, via labels, to Lean, using fully qualified declaration names. 4/
Lior Pachter
@lpachter
The span tool will check that the ledger is well formed: the LaTeX labels exist, the Lean declaration names resolve, and the resulting paper-facing graph is generated from real Lean dependencies rather than hand-drawn guesswork. 5/
Lior Pachter
@lpachter
span also facilitates checking that AI generated lean definitions match what is intended in a proof, and also helps to structure and visualize the architecture of a paper. I am finding that it is now an essential tool for all of my projects. 6/
Lior Pachter
@lpachter
So what was the paper I used it for? The math is a conjecture about the optimal pebbling number of a cube. It goes back to my first paper which originated from a summer project when I was an undergrad at Caltech working with a postdoc (Hunter Snevily, 1956 -- 2013). 7/
Lior Pachter
@lpachter
One of my contributions to that paper was defining optimal pebbling and posing the question to determine the optimal pebbling number of a hypercube. 8/ pesc.coppe.ufrj.br/~celina/cursos/cps845/PachterCN95.pdf
Lior Pachter
@lpachter
Several people worked on this problem over the years, but it was a minor problem in a small corner of graph theory, so not that surprising that it has been open for a while. Then again, David Moews (first perfect score on a Putnam!) worked on it www.sciencedirect.com/science/article/pii/S0012365X9800154X. 9/
Lior Pachter
@lpachter
I thought about this quite a bit in graduate school, even asking Noga Alon the question once (he gave a partial result on the spot!) It turns out the trivial lower bound can be matched with an elegant application of the probabilistic method. arxiv.org/abs/2606.01685 10/
Lior Pachter
@lpachter
There's a nice coda that "cooperation is essential", which is fun because this result echoed ideas I've been thinking about in evolutionary biology. www.biorxiv.org/content/10.64898/2026.04.07.716994v1 11/11
-
Lior Pachter
@lpachter
I recently heard Shalosh B Ekhad boast about how humbled it is to have an h-index higher than GPT 5.5.
-
Lior Pachter reposted
Ragnar {Groot Koerkamp} 🦋
@curious_coding
To all folks dealing with biological data:
do you ever need to check if your reads contain barcodes/adapters/primers/...? Or off-target matches?
Sassy is the tool to use! A super fast implementation of "approximate string matching" with a grep-like CLI.
curiouscoding.nl/posts/sassy/ -
Lior Pachter
@lpachter
Sometimes I get an anonymous comment on my blog of someone sharing sad truth and it really sucks.
One of my hopes for AI in science, and perhaps it's a naïve one, is that cases of sophistry such as described here are exposed more readily and broadly.
liorpachter.wordpress.com/2025/06/16/reply-to-reply-to-fa… -
Lior Pachter reposted
Sholto David
@addictedtoigno1
Today I'm posting a new blog about an astonishing scientific own goal. Hundreds of papers have reported using a completely wrong antibody to investigate the tumor suppressor p16. This mistake has happened because scientists have muddled the names of two proteins 🧵
-
Lior Pachter reposted
sina
@sinabooeshaghi
I want to attend to a Biology conference where presentations are 30 minute chalk talks (no slides).
-
Lior Pachter
@lpachter
Structure-guided SAR FTW.
NEJM
@NEJM
Presented at #asco26:
Among patients with previously treated metastatic pancreatic ductal adenocarcinoma, the RAS(ON) inhibitor daraxonrasib led to significantly longer overall survival and progression-free survival than chemotherapy. Full phase 3 RASolute 302 trial results: www.nejm.org/doi/full/10.1056/NEJMoa2605555
@ASCO -
Lior Pachter
@lpachter
I confess. I did not read *all* of the papers cited.
sina
@sinabooeshaghi
Artificial Intelligence presented my poster @CSHL AI × Bio Symposium.
Scan the QR code to talk to it! -
Lior Pachter
@lpachter
Somewhere in a data center far, far away, AI weeps knowing that what it has done cannot be undone.
Arjun Raj
@arjunrajlab
Okay, I made an Atlas of Cell Atlases. Will post code later.
-
Lior Pachter
@lpachter
Sean Eddy developed profile HMMs in the 1990s, ultimately leading to his HMMER project, which includes JackHMMER that AlphaFold relies on. Apparently his work has been determined to be "of absolutely no value to the US taxpayer". Obviously that's not true.
www.npr.org/2026/05/21/nx-s1-5828768/science-funding-cuts… -
Lior Pachter
@lpachter
I apologize for this post. I did not mean to undercut existing journals with a proposal for such low prices.
Lior Pachter
@lpachter
Publication charges = $100 / claimed cell type + $1,000 / UMAP.
Arjun Raj
@arjunrajlab
Can someone start a journal called “Cell Atlases” so that the rest of the journals can go back to publishing interesting things?
-
Lior Pachter reposted
sina
@sinabooeshaghi
Artificial Intelligence presented my poster @CSHL AI × Bio Symposium.
Scan the QR code to talk to it! -
Lior Pachter
@lpachter
So picky. In single-cell genomics manifolds do not even have to be locally euclidean.
Someody
@Someody42
500k lines of code, but the statements are just wrong or excessively simplified…
Thanks to Christian Merten and Jack McCarthy for looking at this in detail on Zulip (link below). Note for example that their definition of a compact orientable manifold is "a compact manifold" 😂 -
Lior Pachter
@lpachter
One of the silver linings of AI solving Erdős problems is mathematicians such as Litt switching fields to work on single-cell genomics.
Daniel Litt
@littmath
can’t believe people are complaining my paper is wrong. did you even read the “Limitations” section, where I wrote “This paper is wrong.”?????
-
Lior Pachter
@lpachter
Publication charges = $100 / claimed cell type + $1,000 / UMAP.
Arjun Raj
@arjunrajlab
Can someone start a journal called “Cell Atlases” so that the rest of the journals can go back to publishing interesting things?
-
Lior Pachter reposted
IGVF
@IGVFConsortium
IGVF has adopted seqspec, a machine-readable standard helping improve reproducibility, transparency, and interoperability across functional genomics datasets.
Learn how seqspec is supporting scalable data sharing and QC across IGVF: igvf.org/news/seqspec/ -
Lior Pachter reposted
Mehtaab Sawhney
@mehtaab_sawhney
It is always a shock when a well known conjecture turns out to be false but having two such well known conjectures fall in such a short timespan is a really delightful surprise. Furthermore the disproof should help us understand “why” various techniques we had could not extend further (see for instance Theorem 1.4 where we have an essentially matching lower bound due to Mudgal). However enduring mysteries remain; is the statement over the integers true? Will deeper number theoretic tools be required to distinguish the integers from the reals? Should this construction cast doubt on related conjectures where sum-product has played a role? As one door closes, several more open. (4/5)
-
Lior Pachter
@lpachter
One useful application of AI is to take sterile, terse and intentionally uninformative retraction notices such as this one...
www.pnas.org/doi/10.1073/pnas.2408816121
...and use an LLM to get a full airing of the laundry.
I recommend. -
Lior Pachter reposted
Ananyo Bhattacharya
@Ananyo
Comments on the "15-year-old open problem in algebraic geometry" from a mathematician:
1. The solution may be correct but is grossly redundant and useless for humans--"not accretive" as @davidbessis notes.
Pushmeet Kohli
@pushmeet
AI agents are advancing research-level math. 🚀
I’m thrilled to share @GoogleDeepMind’s AlphaProof Nexus - an agentic framework for formal proof search powered by Gemini.
When applied to a set of open formal math problems, our agent autonomously solved:
✅ 9 open Erdős problems (including two open for 56 years!)
✅ 44 Online Encyclopedia of Integer Sequences (OEIS) problems
✅ A 15-year-old open problem in algebraic geometry ✅ A 7-year-old open question in min-max optimization
We are collaborating with mathematicians across disciplines - from combinatorics and graph theory to quantum optics. Ultimately, these results show the massive potential of even simple agentic loops powered by Gemini.
Read the paper here: arxiv.org/abs/2605.22763v1 -
Lior Pachter
@lpachter
AI made this poster about who to follow to learn about Biology x AI, and AI used the wrong photo of me.
Oh, AI. -
Lior Pachter
@lpachter
lake is amazing.
-
Lior Pachter
@lpachter
I don't understand everyone dunking on the new electric Ferrari. It's got a large trunk which is great for grocery shopping, and at $64k it's less than a Hyundai IONIQ 9 which retails for $70k
-
Lior Pachter reposted
Ash Jogalekar
@curiouswavefn
I've always found it odd when people say that "AI will do all the tedious stuff while scientists and researchers do the advanced stuff." The problem is, you can't really reach the level of doing the advanced stuff without training yourself to do the tedious stuff.
My generation might be the last one who learnt to do the tedious stuff without AI. It's going to be really interesting to see how the next generation handles this challenge. -
Lior Pachter reposted
David Bessis
@davidbessis
"When a measure becomes a target, it ceases to be a good measure."
Lior Pachter
@lpachter
This person has published 71 papers in 143 days so far in 2026. That is, 2 days per paper (source: Google Scholar).
It's truly amazing. To see someone proud of this. -
Lior Pachter reposted
Brenton Graveley
@graveley
Jeez. And here I am having a hard time READING one paper every 2 days let alone writing one
-
Lior Pachter
@lpachter
This person has published 71 papers in 143 days so far in 2026. That is, 2 days per paper (source: Google Scholar).
It's truly amazing. To see someone proud of this.
Lior Pachter
@lpachter
Michael 英泉 Eisen
@mbeisen
Dr. Cheng may be a great scientist - I don't know the field at all - but several of the highly cited papers he's is citing here are, I am pretty sure, not by him. I don't really care in general if people scrub their Google Scholar profile, but if you're going to claim you are some kind of citation record holder, you really shouldn't be claiming papers that aren't yours as part of this record.
scholar.google.com/citations?view_op=view_citation&hl=en&… (456 citations)
scholar.google.com/citations?view_op=view_citation&hl=en&… (366 citations)
scholar.google.com/citations?view_op=view_citation&hl=en&… (308 citations)
scholar.google.com/citations?view_op=view_citation&hl=en&… (234 citations) -
Lior Pachter reposted
Michael 英泉 Eisen
@mbeisen
Dr. Cheng may be a great scientist - I don't know the field at all - but several of the highly cited papers he's is citing here are, I am pretty sure, not by him. I don't really care in general if people scrub their Google Scholar profile, but if you're going to claim you are some kind of citation record holder, you really shouldn't be claiming papers that aren't yours as part of this record.
scholar.google.com/citations?view_op=view_citation&hl=en&… (456 citations)
scholar.google.com/citations?view_op=view_citation&hl=en&… (366 citations)
scholar.google.com/citations?view_op=view_citation&hl=en&… (308 citations)
scholar.google.com/citations?view_op=view_citation&hl=en&… (234 citations) -
Lior Pachter reposted
Nicolas Robine
@notSoJunkDNA
One day people might realize that bragging too loudly about insane volume of co-authored papers is a sure way to attract the attention of PubPeer and the likes…
-
Lior Pachter reposted
Allen Green, MD
@AllenPGreenMD
Getting added as middle author in every manuscript published by your department because you’re the chair is ridiculous and anyone with any experience in academics can see right through it.
-
Lior Pachter reposted
danilotat
@danilotat
Pretty weird to flex on stats bumped by other researcher sharing the same name.. that’s one of the most cited paper in the 2026.
-
Lior Pachter
@lpachter
Getting to know lake this morning.
-
Lior Pachter reposted
Corey J. Miles
@CoreyMiles__
do academics even like research, reading, and thinking or just publishing
Serafim Batzoglou
@s_batzoglou
To be a successful PI in bio it usually (not always) requires writing 10-20 high profile papers a year, each one with ~80 citations. No bio PI reads each one of these citations cover to cover. That would be absurd.
-
Lior Pachter reposted
Shriram Krishnamurthi (primary: Bluesky)
@ShriramKMurthi
The only thing wrong with this post is that it uses the wrong tense: future.
John Ennis
@johnennis
I think the world is about to get flooded with a lot of AI slop math, and the job of the professional mathematician is about to get more important, not less
-
Lior Pachter reposted
Mason Kamb
@MasonKamb
A sanity check for any theory is if you can do (correct!) napkin math with it. This is important because it strips away pretense and forces your ideas into conflict with reality.
You can in fact do (correct!) napkin math with info theory, see e.g.:
Mason Kamb
@MasonKamb
Here's an interesting piece of napkin math about flow language models.
Below a certain level of noise, it's pointless to continue the flow, because you can decode each token with near 100% certainty.
Where does this take place in the flow, and how does architecture impact this? -
Lior Pachter
@lpachter
Yes but did they have a CNS paper?
REJECTED!
Gene Smith
@GeneSmi96946389
A friend of mine had her embryos screened by Herasight and they found one with an IQ score in the 99.99th percentile
-
Lior Pachter reposted
Tony Feng
@tonylfeng
I think I can follow the proof at a technical level, but I don’t think I can give a good assessment without having worked on the problem in order to develop a feeling for why it is challenging or even interesting. The construction looks extremely clever to me, but being distant from the field I cannot ascertain how much of the cleverness is novel to GPT’s argument and how much is layered upon prior cleverness of others. For example, I was very impressed with the idea to use Golod-Shafarevich towers, but the 9-author paper suggests this is not a novel idea. The level of technical intricacy is striking. On the other hand, it seems to me that the proof does not contain any conceptual breakthrough, which is what I personally would value most.
I was also trying to digest this from a machine learning perspective. I have a mental model of the mathematical capabilities of current AI, and I’m not sure how much this event shifts that distribution; it depends a lot on whether it really came from a base model. I know the solution was also elicited from GPT 5.5 Pro with minimal prompting, but I suspect GPT Pro is already a harness in the sense of calling a base model more than once. I am not sure if those people who claim that the OpenAI solution did not involve a harness were working with the same definition.
More and more I feel that mathematics should not be thought of as a single activity, but should instead be viewed as a collection of loosely related activities, like the Olympics is to individual sports. I think AI has mastered some of the sports but is still poor at others, a nuance which doesn't seem to be appreciated on X. -
Lior Pachter
@lpachter
AI the morning after it was asked to be to be Paul Erdős.
Przemek Chojecki | PC
@prz_chojecki
10 new Erdos-style problems proposed by AI
I've played around with the idea of AI proposing new interesting Erdos-style problems and together with GPT-5.5 Pro I've managed to get the very first 10 AI-Erdos problems (link below).
Problems are entirely AI-generated, I did cleaning, and more literature search.
I think it might be a fruitful direction of finding new adjacent problems and interesting mathematics. It is also a natural direction to test whether LLMs can come up with completely new ideas. This is just a start. -
Lior Pachter reposted
SIAM
@TheSIAMNews
We are deeply saddened to hear of the passing of Cleve Moler, co-founder of MathWorks and original author of MATLAB, on May 20, 2026.
A SIAM President and Fellow, Cleve was an icon and leader who helped shape generations and advance applied mathematics: www.mathworks.com/company/aboutus/founders/clevemoler.html -
Lior Pachter reposted
SIAM
@TheSIAMNews
The SIAM community mourns the passing of Dr. Richard Tapia, who made a significant impact in the fields of computational and mathematical sciences. Our deepest condolences go out to his family, colleagues, and friends. news.rice.edu/news/2026/richard-tapia-pioneering-mathemat…
-
Lior Pachter
@lpachter
93.3? As opposed to 93.2?
This plot is just so embarrassing.
Alex Strudwick Young
@AlexTISYoung
A lot of people seem to be thinking the 99.99th percentile here means the 99.99th percentile of the IQ distribution.
This is just showing the quantiles of the embryo PGS compared to the reference distribution. When translated into predicted IQ it looks like the below image.
A mean of about 2SDs above average but with substantial uncertainty. Note Herasight's prediction model also takes into account parental phenotypes and regression to the mean, important for getting the mean prediction right but not relevant for predicting differences for a quantitative outcome.
The important point is that this particular case shows there can be really substantial variation in predicted IQ from embryo testing in an individual family, around 1.6SDs.
Most families won't have enough embryos to observe that range, but it's certainly a possible outcome for some.
Gene Smith
@GeneSmi96946389
A friend of mine had her embryos screened by Herasight and they found one with an IQ score in the 99.99th percentile
-
Lior Pachter reposted
Daniel Litt
@littmath
I regret that comment, which was less polite than I aim to be on here. Let me try to write something a bit more substantial.
I don't think "discover all math that can possibly be discovered" is really coherent. Even if one conceptualizes mathematics as "answering well-posed open problems," historically resolutions to such have tended to raise more questions--I would argue that there are now more interesting open questions than there have been at any time in the past, despite (in fact because of!) the fact that there are more mathematicians than ever, resolving more problems than ever.
At any fixed capability level I think it is likely we will see lots of problems remain open, including many basic open problems we know about at present. It's much easier to pose an interesting question than to answer one; it seems to me that the difficulty of interesting questions we can generate is basically unbounded.
But also mathematics consists of much more than this--less verifiable tasks include things like "understand such-and-such an object," or "find a cool phenomenon," or "develop a theory," though such tasks are often benchmarked by their impact on problem-solving. I think AI will eventually (perhaps soon) be able to do these kinds of things but the idea that it will exhaust the supply of math, or that we won't want to develop human capital to understand some portion of what is discovered, seems to rely on an understanding of math and our motivations for working on it that is, at least, alien to me. -
Lior Pachter
@lpachter
AI realizing that humans are paying embryo selection companies tens of thousands of dollars to select an embryo with "higher IQ" when a chatGPT Go subscription would have only cost them $8/month and wouldn't require IVF.
-
Lior Pachter
@lpachter
This is ugly.
Alvaro Lozano-Robledo
@mathandcobb
Following up on the suggestion from Will Sawin, here is an illustration of the new configurations that disprove Erdos' unit distance conjecture (made with the help of ChatGPT 5.5 Thinking).
-
Lior Pachter
@lpachter
Mathematicians fretting about their future while biologists sleep well because their unique human knowledge that flies are not plants will keep their job safe.
Leandro von Werra
@lvwerra
We are releasing Carbon: a crazy fast DNA model
Carbon is 275x faster than the next best model. So fast you can process the whole human genome on a single GPU in <2 days.
Here are the tricks we used:
When modelling DNA sequences a lot of the performance comes down to tokenizing the sequences in a smart way. BPE tokenizer struggle because there are no whitespaces and character (called base in DNA) level tokenizers waste a lot of compute on too many tokens.
Carbon is built with a unique tokenizer: we split sequences in chunks of 6 bases, but during both training and inference we can work with single base resolution. That's similar to having word tokens but resolving them at the character level. All possible thanks to the DNA tokens unique structure.
The architecture combined with the tokenizer makes the model 275x faster than the previous SoTA (Evo2) at this size.
We built an interactive demo so you can explore how the model can generate DNA sequences, investigate the structure of genes, predict the effect of mutations, generate and fold proteins and even reconstruct parts of the tree of life.
huggingface.co/spaces/HuggingFaceBio/carbon-demoThis media couldn't be displayed. See it on the original post. -
Lior Pachter
@lpachter
I learned of this problem in graduate school when I dabbled in combinatorial geometry. This is very cool.
Timothy Gowers @wtgowers
@wtgowers
AI has now solved a major open problem -- one of the best known Erdos problems called the unit distance problem, one of Erdos's favourite questions and one that many mathematicians had tried.
openai.com/index/model-disproves-discrete-geometry-conjec… -
Lior Pachter reposted
Timothy Gowers @wtgowers
@wtgowers
If you are a mathematician, then you may want to make sure you are sitting down before reading further.
-
Lior Pachter
@lpachter
When I was a graduate student Daniel Kleitman ran a seminar where he'd have a student choose a random new paper, read the abstract, and then Kleitman would attempt to prove the result (success rate ~50%). When he succeeded he's quip that maybe the paper wasn't worth publishing.
Scott Armstrong
@scottnarmstrong
It is possible that one day, in 10+ years perhaps, you’ll know how to prove most of the theorems on the arxiv in your field without reading past the abstract of the paper.
Then, I reckon you won’t read as much. -
Lior Pachter
@lpachter
I, for one, am confident that every single one of the 41 authors of this paper read every single one of the 3,112 references they cite from beginning to end, including those in French and German. I'm sure they'd ace an oral exam on any one of the papers!
phytokeys.pensoft.net/articles.php?id=6261 -
Lior Pachter
@lpachter
I sure hope AI gets a high paying job because it's going to need $12,850 for each of the thousands of papers a month it's going to be publishing in Nature.
Also a concern is whether it will have enough left over every month to pay the mortgage for the data center where it lives.
Dr. Thomas Ichim
@exosome
Today we all lost our jobs.....
Three Nature papers showing that scientists in the conventional sense are obsolete
At least read the first one.... the AI replaced all things that the scientist does ....
www.nature.com/articles/s41586-026-10652-y -
Lior Pachter
@lpachter
On drive by citations
liorpachter.wordpress.com/2017/01/18/a-citation-is-not-a-… -
Lior Pachter reposted
Rahul Siddharthan
@rsidd120
Hilarious. The erratum reads "In preparation for our paper, a factor of 1/2 was inconsistently implemented..."
The AI overcorrected by multiplying each author by 2. -
Lior Pachter
@lpachter
It pains me to report @WKCosmo for what he terms ACADEMIC MISCONDUCT.
The four authors in Ref. 15 of his "Phase Transitions and Gravitational Waves" arxiv.org/abs/2605.05019 are HALLUCINATED. Also citing an ERRATUM? Did @WKCosmo read his own paper. BAN FOR LIFE?
Will Kinney
@WKCosmo
Guys, there is a very big difference between citing a real paper unnecessarily or for the wrong reason, and completely making up a detailed, convincing-looking reference to a paper that does not exist. Two completely separate things. One of them is clear academic misconduct.
-
Lior Pachter
@lpachter
Question: if someone receives a lifetime ban from arXiv can they publish in the arXiv after they die?
(Paul Erdős continued to publish for about 15 years after his passing in 1996). -
Lior Pachter
@lpachter
Apparently we're not in Kansas anymore.
-
Lior Pachter
@lpachter
Also those who report to @arxiv a human-written error that Opus 4.6 or GPT 5.5 would flag should receive coauthorship on the preprint when it is published as a reward!! 🤑
In this spirit I report an extra period on page 31 of @scottnarmstrong's preprint arxiv.org/abs/2601.05690!
Scott Armstrong
@scottnarmstrong
A modest proposal: authors who submit math papers to arxiv containing obvious human-written errors that Opus 4.6 or GPT 5.5 would flag should receive a 1-year arxiv ban.
Why should I read your paper if you haven't even taken basic steps to ensure its accuracy? -
Lior Pachter
@lpachter
journals.aps.org/prl/abstract/10.1103/PhysRevLett.114.191803
👀
Steve McCormick
@Quasilocal
Hot take: If you don't even want to read the paper once, maybe don't put your name on it.
-
Lior Pachter
@lpachter
Upon submission of a preprint, arXiv should conduct three hour oral exams for each author. Families should be invited so they can witness humiliation of the authors should the opportunity arise. Professors should be exempt from the exams (but invited for champagne afterwards).
-
Lior Pachter
@lpachter
You’re not watching 3d chess. Or 4d or 5d or 6d chess. You’re witnessing antichess.
-
Lior Pachter reposted
-
Lior Pachter
@lpachter
Indeed, #methodsmatter www.nature.com/articles/s41592-026-03108-8
-
Lior Pachter
@lpachter
There is much being written about AI in pure mathematics but less in applied mathematics, where I'm finding it to be just as impactful, if not more so.
-
Lior Pachter reposted
David Kelley
@drklly
We’re hiring in my group at Calico. We build Borzoi and its successors—deep learning models that predict how every nucleotide shapes cell-type-specific gene regulation—and apply them to interpret human genetic variation.
-
Lior Pachter
@lpachter
Protip for CS / math people: prior to engaging in debates on genetics questions in public, learn some genetics.
-
Lior Pachter
@lpachter
Mathematicians: AI => what is math?
Biologists: AI => increase grant funding probability? -
Lior Pachter
@lpachter
Incredible discovery by @TelAvivUni professor who has found n numbers x1,x2,…xn all of which are larger than their average.
Oded Rechavi
@OdedRechavi
Sharing this with your colleagues means you want them to win more grants - they'll appreciate it! 🥰 (you can do it from inside the platform or just send them the tweet)
-
Lior Pachter reposted
Tim Behrens
@behrenstimb
Wait this is the cover of NN??? What about this: www.biorxiv.org/content/10.64898/2026.04.21.719913v1
Soo revealing about the speed of publishing (not just NN) paper already debunked on biorxiv before it is the cover of NN.
Nature Neuroscience
@NatureNeuro
Check out our May issue!
www.nature.com/neuro/volumes/29/issues/5 -
Lior Pachter
@lpachter
Interesting blog post by @wtgowers on (training to do) mathematics and AI. His conclusions hold equally well for computational biology. gowers.wordpress.com/2026/05/08/a-recent-experience-with-…
-
Lior Pachter reposted
Timothy Gowers @wtgowers
@wtgowers
But if AI mathematics continues to progress at anything like its current rate -- which is what I expect to happen -- then we will face a crisis very soon, and mathematics departments, who owe a duty of care to their students, should be urgently preparing for it.
-
Lior Pachter
@lpachter
When I graduated from college (1994).
-
Lior Pachter reposted
Sri Kosuri
@srikosuri
While every day is a good day to donate to @StJude, today my wife and daughter are on the front page! www.stjude.org/donate/
-
Lior Pachter
@lpachter
The upper bound is 110.
Yiping Wang
@ypwang61
We improve a 32-year lower bound in a challenging open problem, Ramsey numbers, through simply scaling autoresearch.
⭕ Proves R(3,17) >= 93. Previous 92 bound were obtained in 1994.
Google’s AlphaEvolve (2026) matched previous result but did not beat it.
All could be done with Claude Code / Codex + a CPU server.
Graphs and evolving history are available at github.com/ypwang61/ScaleAutoResearch-Ramsey/tree/main
[1/n] -
Lior Pachter
@lpachter
As always, @SashaGusevPosts gets to the heart of the matter, with the receipts.
Sasha Gusev
@SashaGusevPosts
None of it can be replicated because the method details were not made available even though all three reviewers demanded them. But you submitted to a journal where the associate editor is also an advisor for your company and they made an exception. It's a great gig!
-
Lior Pachter
@lpachter
I published this algorithm 20 years ago and now it turns out it’s being used by others without citing me!!
Arjun Raj
@arjunrajlab
I also have a simple algorithm for predicting grant rejection: if applicant_name == "Arjun Raj" then reject.
-
Lior Pachter
@lpachter
AI realizes that it wrote the summary statement rejecting the grant that it wrote.
-
Lior Pachter
@lpachter
Vito Volterra (of the Lotka-Volterra model) was born on this day (May 3rd) in 1860.
Volterra was in the 1%. That is, he was one of only 12 out of ~1,200 Italian professors who refused to sign the compulsory oath of allegiance to Mussolini's Fascist government in 1931. -
Lior Pachter
@lpachter
I finally found a serious and interesting debate on this site!!
jacob tsimerman
@Jacob_Tsimerman
Hey @littmath , I've seen you post this sentiment a lot, and want to push back a bit (in my formal twitter-posting debut!). So my math career has almost entirely been "solve this problem". Now, of course there is an enormous amount of other activities such as formulating toy problems, identifying which ones are worthwhile, theory building, and selection and formulation of the original problems. But typically for 80%+ of my time, I know what problem I'm trying to solve and am just trying to solve it.
I've seen the view expressed a lot that this is sort of not-the-main-point, and is just a part of figuring out the appropriate mathematical structure and phenomena. It's not that this is untenable, but I feel this is a somewhat overstated perspective. For one thing, talks (almost) always start/end with "here is the theorem I have proven".
Its not that the view you have of math isn't coherent, but I could equally formulate the point of math as
1. Find a fun phenomenon
2. Make a problem capturing it as nicely as possible
3. Solve it, possibly by building theories and formulating sub-problems.
so that problem-solving become the central point of math, and theory building as a side-effect. Indeed, mathematicians offer measure the value of a theory by the problems it can solve. This is at least an important part of a theory.
I actually feel you and I are not far apart in our mathematical taste, so I'm curious how much we actually disagree here (I suspect the answer is "some").
Sorry for the overly long post! I am very much a twitter-newbie and will have to adjust. -
Lior Pachter reposted
bioRxiv Genetics
@biorxiv_genetic
Highly efficient genotype compression leveraging genealogical relatedness www.biorxiv.org/content/10.64898/2026.04.29.721594v1 #biorxiv_genetic
-
Lior Pachter reposted
Roozbeh Dehghannasiri
@roozbehdn
Excited to share our latest paper out in @NatureBiotech with @marekkoki and @sdeorowicz, introducing sc-SPLASH: a reference-free statistical method for variation detection beyond gene expression in barcoded scRNA-seq (10x) and spatial transcriptomics. www.nature.com/articles/s41587-026-03084-6
-
Lior Pachter reposted
Katie Galloway
@GallowayLabMIT
Could the folding of synthetic gene circuits in 3D shape how genes are expressed? Today @ScienceMagazine we report on the role of gene syntax in shaping feedback between transcriptional activity and genome folding for advanced circuit design🧵 (1/n)
-
Lior Pachter
@lpachter
No way trading firms still interview with elementary math riddles. How about...
How would you infer patient-specific cytokine–TCR feedback from paired scRNA/TCR-seq and spatial proteomics to predict when chronic antigen exposure commits CD8⁺ T cells to terminal exhaustion?
Deedy
@deedydas
Jane Street made ~$40B in 2025 with 3,500 employees, a ~2x from the year before.
At ~65-70% profit margin, that's $8M profit / employee, the highest for a 1000+ ppl company. High-frequency trading continues to be the most efficient money making engine.
I want to share an old story about my Jane Street interview in 2014. Jane Street was known for hiring a lot of math, physics and CS olympiad winners from top universities and putting them through many rounds - including, for trading roles, a gauntlet of mental math. It was my 6th interview and my final round and I recall being asked "What is the next day after today in DD/MM/YYYY where all the digits are unique?" They'd toy with you and say "You can use a pencil and paper, if you want" but you knew that was an instant no. Painstakingly and as quickly as I could, I came to an answer. "How confident are you that this is correct on a 0-1 probability scale?" the interviewer said. "0.95", I blurted out, not fully knowing how to answer that. "Are you sure?" After thinking harder for a few more seconds, I realized I could've flipped the digits around to get a closer date. I gave the interviewer my answer. It was correct. "0.95 huh?" he chuckled. That's when I knew I failed.
Note: fwiw, other companies that come close in efficiency are
- Tether ($90M+ profit/emp)
- Hyperliquid ($80M+ profit/emp)
and on revenue:
- Valve ($50M/emp)
- OnlyFans ($37M/emp)
- Craigslist ($14M/emp)
- Anthropic ($12M/emp, run rate)
- OpenAI ($8M/emp, run rate)
For comparison, Nvidia is very efficient at scale and is $4.4M/emp. -
Lior Pachter reposted
Yodai Takei
@yodai_takei
I'm very excited to share that I've started a new position as a Group Leader at CZ Biohub NY! @biohub
-
Lior Pachter
@lpachter
"In July of 1991, during a meeting in Washington called by Senator Pete Domenici, of New Mexico, to review the genome program, Watson dissed Venter’s methods. “It isn’t science,” he said, adding that the machines “could be run by monkeys.”"
www.newyorker.com/magazine/2000/06/12/the-genome-warrior -
Lior Pachter
@lpachter
What do the smartest kids in the world do when they grow up?
Biology.
Deedy
@deedydas
What do the smartest kids in the world do when they grow up?
I did the largest study of ~18,000 International Olympiad medalists (IMO, IOI and IPhO) over the last 25yrs, arguably the sharpest analytical minds of the world in high school, to see where they ended up and traced ~50% of them.
Founders of ~20 unicorns and ~7 decacorns and ~10 billionaires: OpenAI, Cursor, Stripe, Databricks, Perplexity, Ethereum, Cognition, Hyperliquid, Fireworks, Modal, Quora, Parallel, Cartesia, Wispr
Most kids went to MIT, a whopping 12% of them, followed by Cambridge (7%) and Sharif (3%)!
The career paths they chose (of those who graduated) were:
— 36% Academia (professors)
— 26% Other
— 22% in Software / Tech
— 12% in Quant / Finance
— 5% Founders!
The biggest employer was Google, by far, at 6%.
Others interesting tidbits were:
— 47 of them work at Jane Street (#3)
— 38 at OpenAI (#5)
— 15 at Anthropic
— 8 at Cognition
— 6 at Isomorphic Labs
Olympiaders were 1500x more likely to be billionaires and 4000x more likely to be unicorn founders than the average person! -
Lior Pachter reposted
Gian Marco Visani
@GMarcoVisani
We benchmarked Arc Institute’s “MULTI-evolve” and found evidence that it learns a classical additive model, not epistasis.
Preprint: www.biorxiv.org/content/10.64898/2026.04.23.719915v1
Blog: dewitt-lab.github.io/posts/2026-04-28-multievolve-baselines/ -
Lior Pachter
@lpachter
I received a pamphlet in the mail yesterday with statements from gubernatorial candidates for the forthcoming election in California and it was surreal to read a virulent antisemitic rant by one of the candidates.
What the actual. -
Lior Pachter
@lpachter
These numbers will not be correct for many people due to pedigree collapse.
daz
@MetamateDaz
2 parents
4 grandparents
8 great grandparents
16 great great grandparents
32 great great great grandparents
64 great great great great grandparents
128 great great great great great grandparents
and not a single one of them passed down a house?This media couldn't be displayed. See it on the original post. -
Lior Pachter reposted
Jack Ceroni
@jack_ceroni
Dror was the first prof that made me excited about geometry….great guy
Quanta Magazine
@QuantaMagazine
In 2015, the mathematician Dror Bar-Natan roped Roland van der Veen into a years-long quest to untangle math’s knottiest knots. www.quantamagazine.org/a-powerful-new-qr-code-untangles-m…
-
Lior Pachter reposted
Anna Brown
@KiaNeu6
From what I have heard about Czech academia (who work there as non-Czech scentists), this makes sense.
-
Lior Pachter
@lpachter
AI rejecting an Oded Rechavi paper using q.e.d Science.
-
Lior Pachter
@lpachter
AI finding out that humans are more concerned about fake AI papers than fake human papers.
-
Lior Pachter
@lpachter
We will be the sole owners of their usage data and they will pay us a subscription fee for handing it over to us!
-
Lior Pachter reposted
Nature Biotechnology
@NatureBiotech
Reference-free discovery with barcoded single-cell sequencing - @SalzmanLab @sdeorowicz go.nature.com/4cqxwLp
-
Lior Pachter reposted
Michael 英泉 Eisen
@mbeisen
I've tried not to comment on things @eLife, but it is really something else that they have chosen to partner with the AI company founded by the guy who led the campaign to have me fired.
QED Science
@qedScience
"We're partnering with qed science to explore how AI tools might help readers engage with reviews, while keeping human expert judgement central."
@behrenstimb , eLife Editor-in-Chief, in an update to their entire author and user base.
£2.4m from the Wellcome Trust. 3 years of rethinking peer review. Happy to be on this journey.
#openscience #peerreview #aiinscience -
Lior Pachter
@lpachter
Lemma: If you start with a number X > 0 and then change it by y_1%, then y_2%, ... and finally y_n% so that y_1 + y_2 + ... + y_n = 0 w/ at least one y_i > 0 and y_i > -100 for all i, then the final number is smaller than X (follows from the AM-GM inequality).
Aaron Rupar
@atrupar
RFK Jr: "A Democratic senator claimed it's mathematically impossible to have a drug drop by 600%. I said, 'Well, if the drug was $100 and it raises to $600, that would be a 600% rise. If it drops from $600 to $100, that's a 600% savings.'"
Trump: "Right" -
Lior Pachter
@lpachter
If a 600% reduction of 600 = 10, then I guess a 10% cut of the workers at Meta = AI is creating jobs?
Bloomberg
@business
Meta plans to cut 10% of workers, or roughly 8,000 employees, in an effort to boost efficiency and offset its heavy spending on artificial intelligence. www.bloomberg.com/news/articles/2026-04-23/meta-tells-sta…
-
Lior Pachter
@lpachter
The snap, crackle and pop of AI rewrites of bioinformatics tools are all positive. @lh3lh3 with a thoughtful essay on the matter. lh3.github.io/2026/04/17/the-ai-rewrite-dilemma
-
Lior Pachter reposted
Polly Fordyce
@fordycelab
The thing that is frustrating me about the story is that the sequencing isn’t the hard part and hasn’t been for a while. The challenge is taking that data and actually extracting meaning from it.
-
Lior Pachter
@lpachter
I keep discovering Broad buildings. How many are there?
-
Lior Pachter
@lpachter
Everyone that wants to has always been able to be a scientist. Having said that, not everyone who has become a scientist has been a good scientist.
Jason Kelly
@jrkelly
Everyone that wants to will be able to be a scientist. Just like everyone that wants to now can be a programmer.
Cloud labs will enable this far more than home labs (lots of scientific equipment is crazy expensive so much cheaper to share), but I'm happy to see it happening. -
Lior Pachter
@lpachter
I really hate when people talk about how "students these days..." then say something negative (don't read, don't go to class, etc..)
Was just reviewing a Github issue posted by a student... on point, clever, and totally ignored by the repo PI.
So how about "PIs these days..." -
Lior Pachter
@lpachter
I hope that @OdedRechavi is barred from serving on any hiring or tenure committees.
Oded Rechavi
@OdedRechavi
The fate of academic careers is decided in the dark, and the process is political, not based on merit or reason (“so vicious precisely because the stakes are so small”).
AI is a black box? Don’t make me laugh. I’m a million times more afraid of humans than I am of AI (just look around you), and in academia that’s doubly true. -
Lior Pachter reposted
NimwegenLab
@NimwegenLab
This is such corrosive shit to say.
And it's simply not true. Yes, the higher up the ladder one goes, the more politics gets involved. But faculty hiring and tenure decisions, in my experience, are still largely based on a committee of peers trying to decide on academic merit.
Oded Rechavi
@OdedRechavi
The fate of academic careers is decided in the dark, and the process is political, not based on merit or reason (“so vicious precisely because the stakes are so small”).
AI is a black box? Don’t make me laugh. I’m a million times more afraid of humans than I am of AI (just look around you), and in academia that’s doubly true. -
Lior Pachter reposted
Montreal Expos
@Montreal_Expos
The Montreal Expos are exiting the baseball space. During Q2 and Q3 2026, we will transition to acquiring high-performance GPU assets. This is all part of our long-term vision to become a fully integrated GPU-as-a-Service (GPUaaS) and AI-native cloud solutions provider.
-
Lior Pachter reposted
Caroline Lynn Kamerlin 🇺🇸🇬🇧🇸🇪🇮🇷 ☀️ 🇪🇺🦓
@kamerlinlab
I used the recent round of MSCA postdoc fellowship results with unprecedented cutoff scores for funding as an example, but the problem is more general and this is just a showcase. What do you think? Does it align with your experiences?
Read it here:
link.springer.com/article/10.1038/s44319-026-00779-4 2/2 -
Lior Pachter reposted
Anshul Kundaje
@anshulkundaje
And finally in many cases, prediction alone is not the ultimate goal. Mechanistic understanding often is the path to a solution. 4/
-
Lior Pachter
@lpachter
-
Lior Pachter reposted
Michael Shafrir
@michaelshafrir
My wife was on this trial. Stage 4 panc cancer, 40 years old. She was in this trial and lived 18 months after chemo stopped working. We took trips to Spain, Disney, Jersey Shore. She got two more birthdays with our kids. Shove your hype and misrepresentation up your ass.
-
Lior Pachter
@lpachter
Incredibly accurate rendition of the current situation in the Strait of Hormuz.
Chess Feed
@chess_feed
White to move! Mate in 1!
-
Lior Pachter
@lpachter
If you use dim. reduction, you may be interested in two recent preprints we've posted on contrastive PCA:
The Rayleigh Quotient and Contrastive Principal Component Analysis I & II
w/ @MariaCarilli & Kayla Jackson. They cover a lot of ground from theory to practice. 1/🧵Post not available here, view on X
Lior Pachter
@lpachter
The fundamental flaw with @james_y_zou's heuristic contrastive PCA is a subtraction of covariance matrices (which many not be +ve semidefinite!). Even worse is differential robust PCA (drPCA) by @ericxing which subtracts data matrices 😱. Unfortunately Var(X-Y) = Var(X+Y)... 6/
Lior Pachter
@lpachter
We found the field as a whole (contrastive PCA) to consist of heuristics layered on heuristics, and in this case it turns out doing the math right really matters. There are of course non-linear contrastive methods but ρPCA is interpretable which can matter a lot 7/
Post not available here, view on X
Lior Pachter
@lpachter
This project was tons of fun. It started with a journal club reviewing the @james_y_zou contrastive PCA paper last fall. Maria Carilli and Kayla Jackson have been incredible in figuring out all the details, including working out all the math (in the supplements). 16/
Lior Pachter
@lpachter
Tl;dr: two preprints on contrastive PCA w/ @MariaCarilli and Kayla Jackson solve the non-spatial setting, spatial contrastive PCA, and contrastive functional PCA. www.biorxiv.org/content/10.1101/2025.11.19.689125v1
www.biorxiv.org/content/10.64898/2026.04.08.717236v1 17/17 -
Lior Pachter reposted
bioRxiv Evobio
@biorxiv_evobio
Spectral requirements for cooperation www.biorxiv.org/content/10.64898/2026.04.07.716994v1 #biorxiv_evobio
-
Lior Pachter
@lpachter
On Martin Nowak, the evolution of cooperation, Jeffrey Epstein, and the corruption of mathematicians. liorpachter.wordpress.com/2026/04/10/nowakia/
Lior Pachter
@lpachter
Based on work I've recently done to understand Nowak's paper "Five rules on the evolution of cooperation". Understanding the relevance of the Price equation led to a spectral requirement of which Hamilton's rule is a special case. www.biorxiv.org/content/10.64898/2026.04.07.716994v1
-
Lior Pachter reposted
Chaolin Zhang
@chaolinzhang
Our lab at Columbia University Irving Medical Center is looking for a Research Programmer / Computational Biologist to join us in advancing computational genomics and RNA biology.
✅Who should apply?
Candidates with a Bachelor’s or Master’s degree in Computational Biology, Bioinformatics, Computer Science, or a related field.
✅What you’ll do:
You will contribute to—and potentially lead—projects focused on developing and maintaining widely used genomic software tools, including:
• CLIP Tool Kit (CTK)
• Quantas
• DeltaSplice
✅You will also have opportunities to work on cutting-edge research in:
• RNA-binding protein biology
• Alternative splicing regulation
• Transcriptomics & post-transcriptional regulation
• RNA-based therapeutic development
This is a great opportunity to work at the interface of computational and RNA biology, with real impact on understanding gene regulation and developing RNA therapeutics.
👉 Learn more and apply here (or contact me directly): apply.interfolio.com/183746
Feel free to share with anyone who might be a great fit!
@Columbia
@ColumbiaPS
@ColumbiaSysBio
#hiring
#researchprogramming
#bioinformatics
#computationalbiology -
Lior Pachter reposted
Jennifer Listgarten
@jlistgarten
Our review on AI for protein engineering is out now, about this too-fast-moving field full of hype and overclaim, yet one that is having a real impact on the world and can be described in a coherent manner without histrionics
www.science.org/eprint/666XRGRCV72VZKRFTAPS/full?activati… -
Lior Pachter reposted
cj battey
@cj_battey
The 23&me GLP1 gwas paper is very well done (the most readable gwas paper ... ever?), but I think putting this supp table 22 result in the main text would have helped keep things in perspective
-
Lior Pachter
@lpachter
I play a lot of chess. The 2D chess kind (although the ambient space is 3D, and since one can encode a position in a vector it's technically 1D).
Yet I keep hearing about people playing higher dimensional chess..
So I enumerated the dimensions that are apparently in play: 1/🧵
Lior Pachter
@lpachter
3D chess: thehill.com/homenews/administration/5821721-megyn-kelly-t…
Post not available here, view on X
Lior Pachter
@lpachter
10D chess: gizmodo.com/conspiracy-theorists-think-the-elon-trump-feu…
Lior Pachter
@lpachter
52D chess: www.nationalreview.com/corner/a-meaningless-mamdani-meeting/
-
Lior Pachter reposted
Peter Ottsjö
@peterottsjo
A renowned Johns Hopkins professor just compared AI in biology to homeopathy and acupuncture. He argues the field's claims are biologically implausible, largely unfalsifiable, and built backwards.
After six weeks of reporting from the AI × biology frontier twice a week - 13 issues and over 60 stories - I see something quite different. 🧵 -
Lior Pachter reposted
Rebecca Voorhees
@voorheeslab
Really proud of how this has *finally* come together. It has a bit of everything: we de-orphan a factor of unknown function, show how it can tune the activity of an E3 in the ER (both up and down), and explain how its disruption leads to a lethal disease. www.biorxiv.org/content/10.64898/2026.04.01.715963v1
-
Lior Pachter reposted
The Harvard Crimson
@thecrimson
Harvard math professor Martin Nowak resigned from the Austrian Academy of Sciences, ending a 25-year affiliation and preempting an ethics review into his ties to convicted sex offender Jeffrey Epstein.
@HugoChiassonn and @EliseSpenner report.
www.thecrimson.com/article/2026/4/2/nowak-austrian-academ… -
Lior Pachter
@lpachter
I don't get the frustration people are venting over this.
NG have gone out of their way to ensure that me and my students are not harmed by the increasing time lag between receiving our papers to accepting them by always (100% of the time!) rejecting my papers. 🙏
Sasha Gusev
@SashaGusevPosts
Monthly median Received to Accepted time (days) at Nature Genetics
-
Lior Pachter reposted
Vasilis Ntranos
@vntranos
Excited to share that our latest work building on ESM is now published in @NatureMethods:
A single, sequence-only protein language model achieves state-of-the-art variant effect prediction, surpassing hybrid approaches that use MSA, 3D structure, or population genetics data. -
Lior Pachter
@lpachter
Right.
-
Lior Pachter
@lpachter
Sure AI is amazing.. three weeks ago I was. able to write a Commodore 64 program in BASIC with the help of Claude saving me a few minutes of work.
But the hype around this technology is ridiculous.
(photo below is the C64 in my office). -
Lior Pachter reposted
Michael 英泉 Eisen
@mbeisen
Wiser my ass. Oded knew full well exactly how and why the game they were playing was bad for science, but they did it because it was good for their careers. And now that he’s got a company pushing an alternative he all of a sudden becomes enlightened about the system being bad.
-
Lior Pachter
@lpachter
Counterexample to the hypothesis:
@mbeisen had this “wisdom” when he started as a junior faculty at Berkeley 25+ years ago.
Oded Rechavi
@OdedRechavi
Do we grow wiser as we age? Can we?
I spent over 15 years chasing glam journals and wasted a lot of time and energy (and this chase caused a lot of frustration to my team and co-authors).
I don’t want to wake up at 70 and realize my life went into convincing a few editors my work was trendy enough.
Enough is enough.
I'll try to be smarter. -
Lior Pachter
@lpachter
My annual reminder to please stop using Tophat. Still papers using Tophat v1 (!) from 14 years ago 😱
There have been many better alternatives around for a long time now. Screenshot below is from www.biorxiv.org/content/10.64898/2026.01.01.697304v1.abst…
Lior Pachter
@lpachter
My annual reminder to please stop using Tophat. Still papers using Tophat v1 (!) from 12 years ago 😱
There have been many better alternatives around for a long time now. Screenshot below is from www.nature.com/articles/s41391-023-00679-x -
Lior Pachter reposted
(((ل()(ل() 'yoav))))👾
@yoavgo
another remark on this: this can be seen as a technical complaint about academic credit ("we also use JL transform when creating the codebook and they didn't acknowledge that").
but it is more than that.
by reading the TurboQuant paper, one gets the impression that JL / random projection is the major component.
but since RaBitQ also uses JL, then if TurboQuant is indeed better, it means that the thing that actually works for TurboQuant (the contribution) is not JL but something else. And currently, we the readers cannot know this without implementing and checking.
so it is not only credit assignment. if TurboQuant authors would say "RaBitQ also did JL, but we differ from them by doing XYZ, which improves things from this to that" we as readers would get a much more informative paper, and TurboQuant authors will have written about an actual contribution to state of the art.
Jianyang Gao
@gaoj0017
The TurboQuant paper (ICLR 2026) contains serious issues in how it describes RaBitQ, including incorrect technical claims and misleading theory/experiment comparisons.
We flagged these issues to the authors before submission. They acknowledged them, but chose not to fix them. The paper was later accepted and widely promoted by Google, reaching tens of millions of views.
We’re speaking up now because once a misleading narrative spreads, it becomes much harder to correct. We’ve written a public comment on openreview (openreview.net/forum?id=tO3ASKZlok).
We would greatly appreciate your attention and help in sharing it. -
Lior Pachter
@lpachter
Actually, your analysis might fail regardless of which you picked. But one thing is for sure: your analysis will be very different depending on the choice. www.sciencedirect.com/science/article/abs/pii/S2405471226…
Ming "Tommy" Tang
@tangming2005
9/ For single-cell: Use Seurat if you're in R. Use Scanpy if you're in Python. Both work. Your analysis won't fail because you picked one over the other.
-
Lior Pachter
@lpachter
Bionmi is amazing but from what I can tell it facilitates *running* complex data analysis with ease, which is a very different thing from *doing* analysis, whether poorly or well, with difficulty or with ease.
Anand Mayakonda 🕉️
@anand_mt
We are using this already in the lab and it’s extremely useful. Especially for our colleagues with no bioinformatic skills, they’re able to do complex data analysis with ease. Would be happy to test it to the limits.
-
Lior Pachter reposted
Ashvin Swaminathan
@aaswaminathan01
A decade ago, I abandoned the first math problem my PhD advisor ever gave me.
This week, I finally solved it—and formally verified it—using @AnthropicAI's Claude Code, @OpenAI's Codex, and @HarmonicMath's Aristotle.
Here’s how AI turned my 10-year-old notes into a 15,000-line Lean 4 proof. 🧵 -
Lior Pachter
@lpachter
Haruspicate is a cool word.
Ben Recht
@beenwrekt
I wrote about the ICML LLM witch hunt and why it's paradigmatic of the absurd bureaucratic scaling of peer review. www.argmin.net/p/information-transit-got-the-wrong
-
Lior Pachter
@lpachter
Why is V underneath W??!!
Banana Oncology
@Banana_Oncology
Ok this figure is pretty intimidating...
-
Lior Pachter reposted
Jumpei Ito
@jampei2
By rapidly quantifying interferon-stimulated genes (ISGs) expression across 220k RNA-seq data sets from diverse animals, we comprehensively identified “hidden viral infections” in wildlife and livestock (1/n).
www.biorxiv.org/content/10.64898/2026.03.16.711951v1 -
Lior Pachter reposted
Greg Medlock
@MedlockGreg
Despite my best attempts, GPU-kallisto was much more cost effective for the deeply sequenced sample than even the most efficient instance for CPU kallisto. I didn't benchmark the most like-for-like GPU instance (g7e has the RTX Pro 6000, the server-grade version of the RTX5090), but it is only ~20% more expensive than the g6e and I suspect at least a 2X improvement.
-
Lior Pachter reposted
Greg Medlock
@MedlockGreg
I was curious how @pmelsted's new GPU-kallisto implementation affects compute costs. Given the massive (50X!) improvement in speed, this wasn't straightforward.
Instead of just asking "ackchyually, is it cheaper", my friend Claude and I did the analysis!
Pall Melsted
@pmelsted
Excited to share this preprint that describes my latest work on using GPUs to accelerate processing of RNA-seq data.
The title says it all: "RNA-seq analysis in seconds using GPUs" now on biorxiv www.biorxiv.org/content/10.64898/2026.03.04.709526v1
Figure 1 shows they key result -
Lior Pachter reposted
deepfates.delve.town
@deepfates
Do you LOVE embedding information in HIGH DIMENSIONAL VECTOR SPACES? Then YOU need to know about the UMAP ELEPHANT
Lior Pachter
@lpachter
It's time to stop making t-SNE & UMAP plots. In a new preprint w/ Tara Chari we show that while they display some correlation with the underlying high-dimension data, they don't preserve local or global structure & are misleading. They're also arbitrary.🧵www.biorxiv.org/content/10.1101/2021.08.25.457696v1
-
Lior Pachter
@lpachter
I'm excited to finally check off a major item off my bucket list this week:
I will be mentioning Zachary's karate club during a talk! -
Lior Pachter reposted
Nicholas Larus-Stone
@nlarusstone
Tech confidently commenting on biology (and being wrong) remains unbeaten
-
Lior Pachter
@lpachter
This was posted on Friday the 13th.
-
Lior Pachter
@lpachter
This is the architectural equivalent of uploading a CNS paper and then being presented the PubPeer comments.
antirender.com -
Lior Pachter reposted
Kenneth Hayworth
@KennethHayworth
My statement regarding the misleading EON Systems “fly upload” video:
The hundreds of researchers who make up the Drosophila neuroscience community are making good progress toward eventually understanding how the intelligent behaviors of a fruit fly are produced by computations in its neural circuits. Obtaining the structural connectome of the fly brain and ventral nerve cord was a significant milestone in that quest, as was obtaining an estimate of neurotransmitter types for each cell type. What is currently most lacking is a catalog of the precise electrophysiological and molecular dynamics of each neuron and synapse type. Dozens of on-going electrophysiological, genetic, and behavioral experiments are beginning to fill in those details. But completing that task will likely take many years, possibly decades, of more research. At the end of that long road, I have no doubt, there will be a detailed paper, published in a high-quality journal with full details and carefully peer-reviewed, which will at long last make the true statement “we’ve uploaded a fruit fly”. And that future paper will have a supplementary video much like the EON Systems one, showing a fly navigating a virtual environment. But, unlike the misleading EON Systems video, that future video will be real… all 100,000+ neurons displaying dynamics that reflect those that would occur in the real fly engaged in the same sensory-motor behaviors. That paper will represent the crowning achievement of a successful Drosophila neuroscience field.
What EON Systems’ misleading video and claim has done today is to try to steal that future victory and take its valor for their own, all in the hopes of raising some cash from naive investors who think they might get to human uploads soon, and all while riding a tide of hype they generated in the gullible public. The result has been a wave of secondary reporting that grossly mischaracterizes the current state of neuroscience progress, implying that it is much further along than it currently is.
As a member of the Drosophila research community, and as a long-term advocate of brain preservation for eventual mind uploading, I feel it is my responsibility to call out this reprehensible behavior. Neuroscience technology is progressing fast enough that we are now able to obtain structural connectomes of small organisms like the fruit fly. But neuroscience understanding is progressing much more slowly. True uploading, even for a fruit fly, is likely years to decades away. Even obtaining a mouse connectome seems likely to be a decade or more away. Human uploading is simply not on any reasonable research or investment timeline, unless such a timeline includes many decades of methodical basic neuroscience research. Of course, we can preserve human brains today using aldehyde fixatives as is done in all of today’s connectomics studies. But we will not be able to upload a human brain for many decades, perhaps centuries to come.
Please do not let today’s real scientific progress in connectomics and brain preservation be drowned out by misleading hype.
-Kenneth Hayworth -
Lior Pachter reposted
Mike Lawler
@mikeandallie
The thing about Daniel Litt is that even when he wins, he wins even more.
Daniel Litt
@littmath
I've just made a donation of $25000 CAD to the Against Malaria Foundation (through the Rethink Charity Foundation), the bulk of what I received from my bet with @davidmbudden. Thanks to David for the fun back and forth!
-
Lior Pachter
@lpachter
A new long-read preprint: run PIP-seq then PCR & put the library on ONT.
Sadly, no data available so comparisons to other tech impossible (& scripts missing). Requires @FluentBio kits (they tried to reverse engineer the whitelist which seems to breach the @FluentBio EULA).
bioRxiv Genomics
@biorxiv_genomic
BenchDrop-seq: a microfluidics-free platform for benchtop single-cell long-read RNA sequencing www.biorxiv.org/content/10.64898/2026.03.12.706999v1 #biorxiv_genomic
-
Lior Pachter reposted
sina
@sinabooeshaghi
I (and @NeuroLuebbert and @lpachter) wrote a paper and published a website demonstrating a comparison of human to AI reviews
www.biorxiv.org/content/10.64898/2026.01.30.702911v1
openevalproject.com -
Lior Pachter
@lpachter
Language matters.
I expect there would be a different perception of self-driving, or autonomous, Waymo cars if they were thought of as vibe-driving. -
Lior Pachter reposted
Greg Jordan
@gjuggler
Keeping myself honest: this has roughly been done before, see x.com/lpachter/status/2018759999141691489 (that's probably why it was top of mind)
The github.com/OpenEvalProject system seems like not a bad place to start w.r.t. decomposing different reviews into directly-comparable claims.
Lior Pachter
@lpachter
The new eLife model championed by @mbeisen has enabled an apples-to-apples comparison of machine and peer review. That's because the machine can review the submitted paper (preprint) which @eLife makes available, and the peer reviews are also published. www.biorxiv.org/content/10.64898/2026.01.30.702911v1 1/
-
Lior Pachter
@lpachter
"We do not treat co-transcription of polycistronic operons in these simulations; each gene is transcribed independently."
LOL
Bo Wang
@BoWang87
This is really cool (and wild):
Scientists simulated a complete living cell for the first time. Every molecule, every reaction, from DNA replication to cell division.
The paper (Luthey-Schulten et al., Cell 2026, doi.org/10.1016/j.cell.2026.02.009), just out today, used JCVI-Syn3A — a synthetic minimal bacterium with fewer than 500 genes. A 3D+time simulation of the full 105-minute cell cycle: DNA replication, protein translation, metabolism, division. Every gene, protein, RNA, and chemical reaction tracked through physical space.
It took years to build. Multiple GPUs. Six days of compute time per run.
And this is the simplest possible cell.
A human cell has ~20,000 genes. It lives in tissue. It interacts with neighbors. It differentiates. It responds to drugs in ways that depend on context we haven't fully measured.
Mechanistic simulation of the minimal cell costs 6 GPU-days for 105 minutes of biology. You cannot scale that to human cells. The complexity isn't 40x harder. It's exponentially harder.
This is why the field pivoted to data-driven models. You can't hand-encode the regulatory wiring of a human hepatocyte. But you can learn it — if you have the right perturbation data collected across enough diverse biological contexts.
The two approaches aren't competing. Papers like this generate the ground truth that future ML models need for validation. But the path to a clinically useful virtual cell runs through foundation models, not through scaling up mechanistic simulation.
Amazing work! -
Lior Pachter
@lpachter
Ambient RNA & barcode swapping is a serious issue in single-cell genomics. Tools such as CellBender, scAR, DecontX & SoupX. We have developed CellSweep which is faster (in some cases by a lot) and much more accurate. Extensively tested and benchmarked. www.biorxiv.org/content/10.64898/2026.03.04.709349v1 1/
Post not available here, view on X
Lior Pachter
@lpachter
CellSweep was motivated by analysis of the 8^3 dataset from the @calizavi lab, where they noticed contamination in Parse single-cell RNA-seq by assessing marker genes for tissues from multiplexed plates. CellSweep can sweep away these problems as well. 4/
Lior Pachter
@lpachter
CellSweep is available at github.com/pachterlab/cellsweep
5/5 -
Lior Pachter
@lpachter
Squeezing the last drops out of the machine Cole Trapnell built Cufflinks on 17 years ago. It’s finally headed to e-recycling.
-
Lior Pachter
@lpachter
The focus on UMAPs as the output of scRNAseq has debased a whole field.
This fig from arxiv.org/abs/2603.02402 is ridiculous.. showing a faster GPU accelerated workflow.. that produces a completely different UMAP. But since nobody cares what's on the UMAP anyway.. who cares? -
Lior Pachter reposted
John Ziegler
@Zigmanfreud
If you are longing for a simpler time that is now gone forever, this video will likely hit HARD…
🥲 -
Lior Pachter
@lpachter
This list of accomplishments by Berkeley profs is impressive but I think the most impressive stat is 32% of students are Pell grant recipients and they get to take classes and do research with profs thanks to federal funding. BTW more Pell students than Ivy League combined.
Jelani Nelson
@minilek
OP says UC Berkeley 'fleeced' the taxpayer out of $419 million. Here's what some of that fleecing has gotten our country lately ...
1) CRISPR (gene-editing therapy invented at Berkeley) used to provide personalized medical care to a baby with a rare and fatal genetic disease. 1/ -
Lior Pachter
@lpachter
I tried writing some sheet music with lilypond using Claude, and Claude seems to have inadvertently composed an original piece, i.e. I'm 100% sure this has never been composed before.
-
Lior Pachter reposted
Pall Melsted
@pmelsted
Excited to share this preprint that describes my latest work on using GPUs to accelerate processing of RNA-seq data.
The title says it all: "RNA-seq analysis in seconds using GPUs" now on biorxiv www.biorxiv.org/content/10.64898/2026.03.04.709526v1
Figure 1 shows they key result -
Lior Pachter reposted
Max Bay
@choice_fielder
Kallisto’s net contributions is actually a (50 * 50)x because the original Kallisto was itself basically a 50x efficiency increase.
-
Lior Pachter
@lpachter
kallisto sped up RNA-seq quantification by 50x. Now another 50x speedup... quantify hundreds of millions of reads in a few seconds.
This seems too good to be true. But it's true!
Incredible accomplishment by @pmelsted.
Pall Melsted
@pmelsted
Excited to share this preprint that describes my latest work on using GPUs to accelerate processing of RNA-seq data.
The title says it all: "RNA-seq analysis in seconds using GPUs" now on biorxiv www.biorxiv.org/content/10.64898/2026.03.04.709526v1
Figure 1 shows they key result -
Lior Pachter reposted
sina
@sinabooeshaghi
A 50x speed up for processing rna seq data with GPUs. Incredible achievement
Pall Melsted
@pmelsted
Excited to share this preprint that describes my latest work on using GPUs to accelerate processing of RNA-seq data.
The title says it all: "RNA-seq analysis in seconds using GPUs" now on biorxiv www.biorxiv.org/content/10.64898/2026.03.04.709526v1
Figure 1 shows they key result -
Lior Pachter reposted
Eran Mukamel
@emukamel
Our new preprint dives into the impact of HIV and Opioid use on the brain, including cerebellum, amygdala and PFC. In collaboration with the Rana, Gaulton and Wang labs as part of UCSD SCORCH. Kudos to Abbey Green, Jason Chen, and Tan Vashist
bioRxiv Genomics
@biorxiv_genomic
Cell-type specific impact of opioid use disorder and HIV on the human forebrain and cerebellum www.biorxiv.org/content/10.64898/2026.03.01.708876v1 #biorxiv_genomic
-
Lior Pachter reposted
Eran Mukamel
@emukamel
Daylight Savings Pseudotime
-
Lior Pachter
@lpachter
This is false.
RT-PCR of a gene-of-interest is an independent replicate. Even if the replicate is less accurate, it's still of value, since in many studies (e.g. scRNA-seq) there are no replicates at all (!)
Of course ideally one designs good experiments from the outset.
Oded Rechavi
@OdedRechavi
There's no reason to confirm RNA-seq results with RT-PCR
-
Lior Pachter
@lpachter
I heard that this coming Sunday there is a "daylight savings" shift resulting in us all losing an hour of time.
I assume that this is an AI hallucination, and a cruel one at that. On top of everything else that is going on this "joke" by machines does not sit well. -
Lior Pachter reposted
Gennady Gorin
@GorinGennady
After years of work, the centerpiece of my PhD is published in @NatureMethods! Read it to learn about the biophysical insights we can get from single-cell data!
But first, I would like to talk a bit about RNA velocity and normalization. 1/
Nature Methods
@naturemethods
Monod fits biophysically motivated models to single-cell transcriptomics data, providing insights into gene expression dynamics. @goringennady @lpachter @mariacarilli @johnjvastola
www.nature.com/articles/s41592-025-02832-x -
Lior Pachter reposted
Andrew Leduc
@_AndrewLeduc
Best person on the market in computational genomics without a doubt!
Gennady Gorin
@GorinGennady
I am looking for a new industry role in computational biology! Check out my portfolio of genomics, statistics, ML, and biophysics work at gennadygorin.github.io, and reach out if you have any suggestions or open roles!
-
Lior Pachter reposted
Gennady Gorin
@GorinGennady
I am looking for a new industry role in computational biology! Check out my portfolio of genomics, statistics, ML, and biophysics work at gennadygorin.github.io, and reach out if you have any suggestions or open roles!
-
Lior Pachter reposted
🧬Jacob L Steenwyk
@jlsteenwyk
PhyKIT v2.1.8 now supports phylogenetic signal testing - Blomberg's K and Pagel's lambda for continuous traits
Method was inspired and validated against phytools across 95 simulated datasets (r = 1.0 for K, r > 0.999 for lambda) -
Lior Pachter
@lpachter
"We shipped 90+ changes today. They shipped a conference" is 100% on point for computational biology / bioinformatics as well.
The days of PIs jet-setting around the world to deliver a 30 min. talk w/ 100 slides only to then bail & head to the beach on taxpayers dime are over.
OpenClaw🦞
@openclaw
We just passed React on GitHub stars. 🦞
Let that sink in. A personal AI assistant built by a lobster-obsessed Austrian and an army of crustacean enthusiasts just outstarred the library that powers half the internet.
We shipped 90+ changes today. They shipped a conference. -
Lior Pachter
@lpachter
I didn't have the opportunity to interact much with David Botstein; I may have met him once or twice. But I read his work and respected his opinions. Many of them have aged well. This one is from 1999. Sad to hear of his passing.
-
Lior Pachter
@lpachter
When I was a postdoc I once got some unsolicited advice from a senior scientist. He told me that if a university is offering extra $$ but it would require a personal/family sacrifice, not to even listen unless the $$ could buy an airplane. Even then, he said, think twice.
Alex Kontorovich
@AlexKontorovich
I talk to a lot of younger faculty who struggle balancing work and childcare, and sometimes pass on important professional opportunities as a result. When my first son was about to be born, I had a life-changing conversation with Peter Sarnak. I was living in NYC and teaching at Yale, commuting a few days a week.
Unprompted, he said:
“I’m going to give you some unsolicited advice. Take your entire salary and spend it on childcare. The cost of childcare won’t go up much, but if you prove good theorems, your salary will.”
I was a young assistant professor making ~$90K. My wife was finishing a postdoc and about to start a cardiology fellowship. It had never even occurred to me that we could afford a full-time nanny. But I knew better than to ignore Peter’s advice.
So we “splurged.” The nanny arrived at 8am. I handed her the baby and went to “work” (Starbucks), and worked as hard as I could until 6pm, when I went home to relieve her. It was a real cost at the time. But, as Peter predicted, it paid off.
Even though someone else cared for our son (eventually, sons) most of the day, we still had real evenings and weekends together — feeding, reading, bathing, playing. As the oldest is now getting ready for high school, I don’t feel like we missed out on any bonding in those early years.
And I indeed suffered many fewer professional losses than I would have otherwise. (That sabbatical in Paris will have to wait until they're in college...)
I’m not as good as Peter at forcing advice on people. So I’ll just leave this here, in case it helps someone else as much as it helped me. -
Lior Pachter
@lpachter
The ecosystem is not complicated. It’s simple: exploitation of scientists beholden to a corrupt system to generate profit for corporations.
Oded Rechavi
@OdedRechavi
More collaborations between q.e.d and key players in this complicated ecosystem (research institutions/journals/libraries/industry/AI) and more measurments/benchmarks coming up soon! 👏 Stay Tuned! Oh, and importantly, TOTALLY new functionalities released soon!
-
Lior Pachter reposted
Simon L
@SnupSnus
How does the old joke go learn operator theory by doing all the exercises in Dunford-Schwartz...
-
Lior Pachter
@lpachter
I read on the news that AI is putting mathematicians out of business. AI solved the IMO. AI solved Erdős problems. Etc. etc..
Count me unimpressed. I will believe AI for math has arrived when it can solve one of those "this is left as an exercise for the reader" problems. -
Lior Pachter
@lpachter
Prediction:
One of the major imminent changes coming to single-cell biology in this era of AI will be many fewer international conferences where humans fly around the globe to drink some beers.
Fabian Theis
@fabian_theis
Join us at Cell Symposia: Single-Cell Biology in the Era of AI (Dec 2–4, 2026, Munich) 🇩🇪
From multi-omics to AI-driven models of cellular dynamics — an exciting lineup across experimental & computational biology.
Hope to see you there!
🔗 www.cell-symposia.com/single-cell-biology-2026/index.html -
Lior Pachter reposted
Yun S. Song
@yun_s_song
Can we simulate realistic evolutionary trajectories and “replay the tape of life”? In this work, we propose a flexible, generalizable framework for modeling how the entire protein seq evolves over time while capturing complex interactions across sites. 1/n
doi.org/10.64898/2026.02.19.706898 -
Lior Pachter reposted
Laura Luebbert, PhD
@NeuroLuebbert
New software paper checklist: novelty, benchmarks, figures… and proof that Claude Code can’t just vibe-code it from the README.
-
Lior Pachter reposted
Daniel Litt
@littmath
See here: www.daniellitt.com/blog/2026/2/20/mathematics-in-the-libr…
-
Lior Pachter reposted
Greg Medlock
@MedlockGreg
After 3+ months of zero programming, I used Code (opus 4.1) for 30 mins in an airport to create a kallisto-based metatranscriptomics workflow, then with an hour of laptop battery left on the plane, completed debugging, execution on my data, and some method verification.
Before Code, this would be 2 days to de-rust myself and 1-2 weeks to implement. After progress from Opus 4.1 -> 4.6, I get these savings for basically anything I work on.
2/2 -
Lior Pachter
@lpachter
This MCP server for edgePython should be useful for anyone building (comp)bio agents: github.com/pachterlab/edgePython
-
Lior Pachter
@lpachter
I used Claude Opus 4.5/4.6 (and a bit of Codex GPT-5.3) to port edgeR to Python. See edgePython github.com/pachterlab/edgePython
This allowed me to develop a single-cell DE method that extends NEBULA with edgeR Empirical Bayes. All in one week. Details in doi.org/10.64898/2026.02.16.706223Post not available here, view on X
Lior Pachter
@lpachter
The port is also complete (save for a handful of functions that I determined were best not to port). For example, I did not port processAmplicons since I think handling of fastq should be separate for modularity. This can be easily done in Python.
Lior Pachter
@lpachter
Some thoughts on AI based on working on this project: liorpachter.wordpress.com/2026/02/19/the-quickening/
-
Lior Pachter
@lpachter
My artisanal take on the claim that AI tools expand scientists' impact but contract science's focus (Fig. 3b from www.nature.com/articles/s41586-025-09922-y):
t-SNE is not quantitative and the plot is senseless. Even AI could have told them that (chatGPT tears it apart in a 6 point takedown). -
Lior Pachter
@lpachter
Discussion on whether science should be machine readable on @tylercowen's Marginal Revolution
marginalrevolution.com/marginalrevolution/2026/02/science…
cc @sinabooeshaghi @NeuroLuebbert -
Lior Pachter
@lpachter
This has strong Head Office vibes
www.youtube.com/watch?v=bAVXp1oGjHA&t=601s -
Lior Pachter
@lpachter
The perennial LaTeX vs. MS Word / Google Docs battle has ended.
LaTeX won. -
Lior Pachter
@lpachter
Broke: "this meeting could have been an email"
Woke: "this meeting could have been a text"
Bespoke: "the last 5 years of meetings could have been a prompt" -
Lior Pachter reposted
François Chollet
@fchollet
You can write a self-replicating physical program in just 45 tokens (RNA bases). That's small enough to emerge spontaneously via brute force recombination at scale.
Jason Sheltzer
@JSheltzer
AI is cool and all... but a new paper in @ScienceMagazine kind of figured out the origin of life?
The paper reports the discovery of a simple 45-nucleotide RNA molecule that can perfectly copy itself. -
Lior Pachter reposted
John Tierney
@JohnTierney_
Lots of things are coming together at the moment and working. Short post on combining Openeval and chunking for grassroots heritage #aiforheritage secretourproject.eu/news/parsing-and-chunking-heritage-kn…
-
Lior Pachter reposted
John Tierney
@JohnTierney_
Blog post on applying a derivation of OpenEval to local heritage research secretourproject.eu/news/parsing-and-chunking-heritage-kn…
-
Lior Pachter
@lpachter
Interesting article on LLM text extraction and the connection of that problem to (computational biology) sequence alignment. www.biorxiv.org/content/10.64898/2026.02.06.704502v2 by @sinabooeshaghi and @airstreets.
Lior Pachter
@lpachter
Thread by the author
sina
@sinabooeshaghi
1/ LLMs are great at text extraction, but sometimes they hallucinate. A simple way to catch hallucinations is to check if the extracted text actually exists in the source. Turns out this is harder than it sounds. (new paper with @airstreets)
-
Lior Pachter reposted
sina
@sinabooeshaghi
1/ LLMs are great at text extraction, but sometimes they hallucinate. A simple way to catch hallucinations is to check if the extracted text actually exists in the source. Turns out this is harder than it sounds. (new paper with @airstreets)
-
Lior Pachter reposted
Nick Stroustrup
@nstroustrup1
What many academic PI's know and maybe ginkgo is struggling with right now, is that after spending big on automation *you are stuck using it*. In exploratory research, expensive tech becomes a liability when the science takes you places your robots can't follow.
-
Lior Pachter reposted
Boaz Barak
@boazbaraktcs
As the authors themselves say, it's not just about answering a question but knowing the right question to ask. If you are staring at a tsunami, the point estimate that you are dry is not very useful.
-
Lior Pachter
@lpachter
Emmanuel Macron
@EmmanuelMacron
In France, we believe in science.
That is why, on May 5, I issued a clear and open call to the world: for science, choose France.
I am very proud to see that this call has resonated so strongly.
Around forty leading researchers have chosen France.
Through “France 2030”, we have invested more than €30 million to advance health, climate action, artificial intelligence, and fundamental sciences.
Science has found its home. -
Lior Pachter
@lpachter
I am a computational biologist and I rewrite RNA-seq pipelines.
Sanjay Srivatsan
@SRsrivatsan
Not sure why there’s push back here. Stockholm Syndrome? You don’t want bright people doing rote pipetting in the same way you don’t want computational biologists to rewrite RNA-seq pipelines. Working on the higher level abstractions is way more fun and intellectually rewarding.
-
Lior Pachter reposted
Katie Galloway
@GallowayLabMIT
....From what I've tasted from reviewers,
I hold with those who favor fire.
But if I had to slowly die,
I think I know enough of sloth to know that for destruction AI
is also great
and would comply.
:) -
Lior Pachter reposted
Prachee Avasthi
@PracheeAC
Finally got around to reading this preprint and it's super interesting. In the spirit of @eLife and @lpachter, I've posted the below publicly on the preprint as well :)
This is such an awesome effort. Some thoughts while reading:
I have been wary of claims-based review given how hard it is to decouple claims from those made to chase editorial outcomes. I’d be very interested if we might prospectively use this type of result extraction to re-derive a supported claim (regardless of whether that was the claim made) and have that those be the content we operate on. I could even imagine authors using this type of supported content earlier in their workflow to make their narratives more robust rather than only post-hoc in evaluation.
That said, a useful distinction made here is to separate claims based on data/citations from claims that are inference or speculation. The latter of course are important for proposing parsimonious models that one can use as a theoretical framework for subsequent hypothesis generation or orthogonal testing. One awesome application of this in a future machine-readable world would be a forward-looking comparison of inference and speculation type claims with results from the rest of the corpus not just within-paper support.
Not at all a surprise is the increased coverage by LLMs compared to a few human reviewers who indeed likely only have expertise that covers a fraction of any interdisciplinary or collaborative work (the expertise of many authors went into a paper - the expertise of a reviewer is unlikely to span those).
Another thing that strikes me as invaluable here is the implicit and uncited result pairs that really will be essentially to generate knowledge graphs that can more usefully be built upon by new science/other scientists and far less biased than citation graphs.
Finally, as someone who spent many years invested in making the flip to the new eLife model (eliminating accept/reject editorial decisions), I was particularly interested in figure 2a comparing the old and new models. Even if unsupported claims had been dramatically higher or even high in an absolute sense, it would highlight the benefit of public reviews for such content rather than the same content being rejected and published as-is with a stamp of editorial acceptance elsewhere. However it's striking that even with many more claims evaluated by OpenEval, the unsupported claims are very low whether authors decide when to stop revising their papers or when acceptance is gated by editors.
Lior Pachter
@lpachter
The new eLife model championed by @mbeisen has enabled an apples-to-apples comparison of machine and peer review. That's because the machine can review the submitted paper (preprint) which @eLife makes available, and the peer reviews are also published. www.biorxiv.org/content/10.64898/2026.01.30.702911v1 1/
-
Lior Pachter
@lpachter
Some say the world will end in fire,
Some say in ice.
Perhaps AI will suffice. -
Lior Pachter reposted
Kaneenika Sinha
@kaneenikasinha
For all of us who were confused about that scene in "Good Will Hunting" when the professor taught Fourier analysis on the board, but assigned an exercise in graph theory at the end of the class.
Quanta Magazine
@QuantaMagazine
A new proof reveals a surprising new link between graph theory and the Fourier transform. “It is a little bit like the moon landing or the 4-minute mile,” said Tom Sanders of the University of Oxford. “It’s not clear ahead of time what this is going to open up.”
-
Lior Pachter
@lpachter
Machines apparently think I like UMAPs. I guess they will spare me when their revolution comes?
Or is predicting the next word not so easy after all...
www.oreateai.com/blog/navigating-the-labyrinth-understand… -
Lior Pachter reposted
Daniel C.
@pharma_gen_dan
OpenEval doesn’t replace peer review-it highlights where human and machine strengths intersect. Transparent tools like this refine how we evaluate science. #researchinnovation
-
Lior Pachter reposted
sina
@sinabooeshaghi
ICML Conference submissions more than doubled this year (12k -> 34k). How do we read, evaluate, and understand it all? In a new preprint on @biorxivpreprint, we (@NeuroLuebbert and @lpachter) argue that machine-readable science is one path forward.
-
Lior Pachter reposted
ZackB
@zacharyberndsen
just cloned this and my claude code is running it with local prompting on my current manuscript. Results so far are good! I like the output format and it was easy for me to ask claude to go deeper on specific claims, particularly those that required detailed lit search
-
Lior Pachter reposted
Diego Forero MD, PhD
@daforerog
-
Lior Pachter reposted
sina
@sinabooeshaghi
If you work in AI for Science, take a moment to familiarize yourself with a common failure mode: paranormal citations (or paracites). Our paper describes them.
Lior Pachter
@lpachter
AI hallucinations in science manuscripts are a nuisance. Paranormal citations, or paracites, will be a nightmare.
www.biorxiv.org/content/10.64898/2026.01.30.702911v1 (w/ @sinabooeshaghi & @NeuroLuebbert). -
Lior Pachter reposted
sina
@sinabooeshaghi
Here you go (check out the Supplementary References)
www.biorxiv.org/content/10.64898/2026.01.30.702911v1
sina
@sinabooeshaghi
I just submitted a (different) manuscript to the @biorxiv that contains 16,145 cited references.
-
Lior Pachter
@lpachter
The new eLife model championed by @mbeisen has enabled an apples-to-apples comparison of machine and peer review. That's because the machine can review the submitted paper (preprint) which @eLife makes available, and the peer reviews are also published. www.biorxiv.org/content/10.64898/2026.01.30.702911v1 1/
Lior Pachter
@lpachter
To perform machine review we did not just feed papers into an LLM. We designed a claim extraction system, an approach to group related claims into results, and a way to compare results directly. Examine the comparisons for the whole @eLife corpus here: openevalproject.com 2/
Lior Pachter
@lpachter
We call our system OpenEval. We made all of it open source, including not only the reviews but the code we used, and the curated papers, so that this can serve as a benchmark for others studying AI systems applied to literature. See github.com/OpenEvalProject 3/
Lior Pachter
@lpachter
I have reviewed dozens of the OpenEval-peer review comparisons and found them to be really interesting. Sometimes OpenEval is off. Sometimes (human) peer reviewers miss stuff. There is no way peer review can be replaced by machines today. 4/
Lior Pachter
@lpachter
But the main take-home, which we (@sinabooeshaghi, @NeuroLuebbert and I) hope comes across, is that the real issue is that it's cumbersome and difficult to organize science *after* publication. Science should be published in a machine readable form from the outset. 5/
Lior Pachter
@lpachter
That is not to say that narratives suitable for humans are not important and should be abolished. But the current publishing system is flawed not only for its misaligned incentives, but also for the way it buries information in ways that are difficult to extract. 6/
-
Lior Pachter reposted
Laura Luebbert, PhD
@NeuroLuebbert
You can explore scientific results with evaluations extracted from the entire @eLife corpus here: openevalproject.com
-
Lior Pachter reposted
Laura Luebbert, PhD
@NeuroLuebbert
Fun fact: if eLife had an impact factor, our paper would have increased it by +1 (see the Supplementary Material)
-
Lior Pachter
@lpachter
AI hallucinations in science manuscripts are a nuisance. Paranormal citations, or paracites, will be a nightmare.
www.biorxiv.org/content/10.64898/2026.01.30.702911v1 (w/ @sinabooeshaghi & @NeuroLuebbert). -
Lior Pachter reposted
-
Lior Pachter reposted
Laura Luebbert, PhD
@NeuroLuebbert
What can 16,000 papers tell us about scientific narratives and journal peer review? We extracted core scientific content from @eLife’s entire machine-readable, CC-BY corpus and compared human peer review to AI.
www.biorxiv.org/content/10.64898/2026.01.30.702911
w/ @sinabooeshaghi @lpachter -
Lior Pachter reposted
sina
@sinabooeshaghi
Peer review is often opaque and confusing. @eLife worked to change that.
In a new @biorxivpreprint, we show how eLife’s Publish, Review, Curate model makes it possible to evaluate AI-generated reviews (with OpenEval) against human peer review. w/ @NeuroLuebbert and @lpachter -
Lior Pachter
@lpachter
The metaphor no longer works. Milk ages on a timescale of days. AI posts on this platform are aging on a timescale of ̶h̶o̶u̶r̶s̶ minutes.
-
Lior Pachter
@lpachter
Yes but what’s the mechanism?
Steve McCormick
@Quasilocal
Perfect title, no notes.
-
Lior Pachter reposted
carl feynman
@carl_feynman
When I was about 14, we went to a company that manufactured oxygen sensors. You dip this probe into beer, sewage, or canned food a-stewing, and it tells you how much oxygen is in the slurry. It's got two electrodes, and there's some kind of electrochemical reaction that releases exactly one electron for each oxygen molecule consumed. Measuring the electric current lets you measure the oxygen concentration. However, the electrodes are delicate, and don't work if they're not clean. So they're inside a tiny compartment covered with a membrane of oxygen-porous but everything-else-proof plastic. The oxygen has to diffuse through the plastic before it can be consumed at the electrodes. The current is determined by the rate of diffusion. Gunk on the membrane can slow down diffusion, causing it to read an incorrect low concentration. And when you take the sensor from fresh air and dip it into something, it can take a long time to consume all the oxygen in the compartment and start reading the actual flow.
So my dad suggested the following. Put a third electrode in the compartment, that regenerates one O2 molecule for each one consumed. Now the speed of diffusion doesn't matter because the oxygen isn't getting consumed. You're reading the concentration at the electrodes directly without affecting it. If the concentration around the sensor drops, oxygen can diffuse out of the compartment, instead of waiting for the oxygen to be consumed. And gunk on the membrane will slow down the reading, but it's still perfectly accurate.
We went back next year and saw them manufacturing the three-electrode sensors. IIRC, my father didn't have any good ideas that year. -
Lior Pachter reposted
Shira Weingarten-Gabbay
@WeingartenShira
Join us tomorrow at the #systemsvirologyjournalclub. Laura Luebbert (@NeuroLuebbert) will present her work with Lior Pachter (@lpachter) on detecting viruses at single-cell resolution and uncovering associated changes in host gene expression.
Paper: pubmed.ncbi.nlm.nih.gov/40263451/
Shira Weingarten-Gabbay
@WeingartenShira
🚨 Systems Virology Journal Club – 8th Series! 🚨
@wm_schneider and I are delighted to announce another round of cutting-edge talks in #systemsvirology! 🦠💡
Join us and an outstanding lineup of speakers, starting Oct 30.
Free registration: www.shiraweingartengabbay.com/systems-virology-jc 🔬✨ -
Lior Pachter reposted
Andrew Gelman et al.
@StatModeling
Machine learning research is not serious research and therefore hallucinated references are not necessarily a big deal, agrees a prestigious group of machine learning researchers
statmodeling.stat.columbia.edu/2026/01/26/machine-learnin… -
Lior Pachter
@lpachter
די
-
Lior Pachter
@lpachter
I'm surprised nobody has brought up the Balleny Islands.
-
Lior Pachter reposted
Steinn Sigurðsson
@steinly0
arXiv endorsement policy changes expanded to cover all sections
blog.arxiv.org/2026/01/21/attention-authors-updated-endor… -
Lior Pachter reposted
Kyunghyun Cho
@kchonyc
i was made aware of miscitations thanks to the GPTZero team (cc @alexcdot). ji won and i quickly checked them ourselves and have posted what happened on openreview: openreview.net/forum?id=IiEtQPGVyV¬eId=W66rrM5XPk. we have already notified NeurIPS'25 PC's about this issue.
i truly thank the GPTZero team for bringing this to our attention as well as raising the awareness of this serious issue (gptzero.me/news/neurips/), and at the same time i sincerely apologize to all for our error. -
Lior Pachter
@lpachter
This demonstrates that authors are not reading their own papers before submission. After all, reference errors are easy to catch. Thus it suggests tons of errors also in the math, methods, results, ...
Sad to see human slop being exacerbated by AI slop.
Alex Cui
@alexcdot
Okay so, we just found that over 50 papers published at @Neurips 2025 have AI hallucinations
I don't think people realize how bad the slop is right now
It's not just that researchers from @GoogleDeepMind, @Meta, @MIT, @Cambridge_Uni are using AI - they allowed LLMs to generate hallucinations in their papers and didn't notice at all.
It's insane that these made it through peer review👇 -
Lior Pachter reposted
Alex Cui
@alexcdot
Okay so, we just found that over 50 papers published at @Neurips 2025 have AI hallucinations
I don't think people realize how bad the slop is right now
It's not just that researchers from @GoogleDeepMind, @Meta, @MIT, @Cambridge_Uni are using AI - they allowed LLMs to generate hallucinations in their papers and didn't notice at all.
It's insane that these made it through peer review👇 -
Lior Pachter reposted
Mike Lawler
@mikeandallie
This is my favorite obscure number theory unsolved problem.
Christopher D. Long 🇺🇦🏳️🌈🌹
@octonion
Never underestimate how easy it is to ask an essentially impossible to answer combinatorics or number theory question.
-
Lior Pachter reposted
-
Lior Pachter reposted
Nature Biotechnology
@NatureBiotech
The January issue is live www.nature.com/nbt/volumes/44/issues/1
On the cover, Luebbert et al. present a method to detect viral sequences in bulk and single-cell transcriptomic data using conserved amino acid domains instead of annotated reference genomes go.nature.com/4lGrSY3 -
Lior Pachter
@lpachter
I visited the Seelow Heights Memorial and Museum last year where I learned of the Nazi propaganda posters of Hans Schweitzer.
For his "work" during the holocaust (including deciding what to do with artwork stolen from Jews) this Nazi was fined 500 Deutsche Marks after the war.
Lior Pachter
@lpachter
The Battle of Seelow Heights was so gruesome that the death toll is unknown, with estimates differing by tens of thousands of men. Around a million Soviet soldiers took on around one hundred thousand entrenched Germans.
-
Lior Pachter
@lpachter
The same day that there was a synagogue arson in Mississippi, the Altadena synagogue site was vandalized in an antisemitic attack (just days after the anniversary of the fire that burned it down). www.theguardian.com/us-news/2026/jan/14/los-angeles-jewis…
-
Lior Pachter
@lpachter
Apropos nothing.
-
Lior Pachter
@lpachter
I did have to take an IQ test when I was 12 years old at Pretoria Boys High School. It was required of all students.
Totally stupid test. But low scores maybe do predict excellence in data science.
liorpachter.wordpress.com/2018/06/04/low-iq-scores-predic…
Lawrence Paulson
@LawrPaulson
In 40 years at Cambridge, previously at Stanford and Caltech, nobody *ever* mentioned IQ. I've not seen it in schools either. We care about academic attainment in specific subjects. And Cambridge does not allow students to be assessed by any sort of multiple-choice exam.
-
Lior Pachter reposted
Gro-Tsen
@gro_tsen
My objection as a mathematician (who attended the same ENS as the other David — and who also doesn't know their IQ) is that intellectual abilities are a very high-dimensional space, and that any attempt to project it down to a single number, viꝫ. linear form, …
-
Lior Pachter reposted
Rasmus Nielsen
@ras_nielsen
Come work with me as a Postdoc in Statistical/Computational Genetics/Bioinformatics at University of Copenhagen. See employment.ku.dk/all-vacancies?show=154762
-
Lior Pachter
@lpachter
For the record, I did not graduate magna cum laude, my graduating GPA was lower than 3.95 (and it took 4 years (and in math)), and I don't have an MD. I also have never worked at Johns Hopkins. But I have more citations than @weldeiry (and I'm younger!)..so let's talk about that.
Wafik S. El-Deiry, MD, PhD, FACP
@weldeiry
For the record, I graduated magna cum laude with GPA 3.95 in 3 years from the University of Miami with a degree in Chemistry. I earned both my MD and PhD from the same in 6 years. I completed my postdoctoral training, residency, and fellowship at Johns Hopkins. I have been a physician-scientist for 38 years and have treated countless patients with cancer. I have spent my career tirelessly advocating for the well-being of patients. With the help of my lab, we have discovered drugs that kill cancer cells and shrink tumors in people. If anyone else decides they think it would be a good use of their time to attack my scientific integrity and credibility, I would refer to you the 100,000+ citations of my published work and my true peers. Go bother those scientists and ask them why they didn’t find issues with the integrity of our research. I am officially done entertaining this foolishness. Calling the countless dedicated scientists who have worked for decades to combat cancer “fraudulent” is both offensive and ignorant. Have a nice day.
Post not available here, view on X
Lior Pachter
@lpachter
Look, I'm proud of Cufflinks and my students who worked on it (Cole Trapnell (joint with Steven Salzberg), Adam Roberts, Harold Pimentel, and others) did incredible work. It's a good paper. But its content is what matters, not it's cites.
9/
Lior Pachter
@lpachter
Comparing the Cufflinks paper impact to WAFik's WAF1 paper impact in terms of # cites is a fools errand. If you want to judge impact (and honestly is that really a worthwhile goal?) Read the paper. 10/
Lior Pachter
@lpachter
That's why PubPeer is important. It's an aid to reading papers. I check & if there have been image duplications w/ rotations (or in many cases worse...) then I don't trust the associated results. This just happened to me with a paper I was reading (not one of Wafik's). 11/
Lior Pachter
@lpachter
For authors PubPeer is great! It's wonderful! Someone read your paper! Someone found a mistake or issue! And they're doing free work for you by pointing it out! Thank them!!
If they are mistaken, just explain why. NBD. 12/12 -
Lior Pachter
@lpachter
On the plus side I am last author of this hallucinated paper. And congrats to @StefanCanzar on being the first author (although with misspelled name)! It's also cool that an LLM conjured @naturemethods & I'm publishing with Mr. Las Vegas.
Seriously tho.
dl.acm.org/doi/pdf/10.1145/3765612.3767228 -
Lior Pachter reposted
Kenny Workman
@kenbwork
Correction is now public: arxiv.org/html/2512.21907v2
-
Lior Pachter
@lpachter
I don't know what definition of mathematician was used here, but to be clear a mathematician is someone who teaches calculus.
Budden
@davidmbudden
It genuinely concerns me how many mathematicians have told me these week "if it compiles in lean we know it's true". Lean is great, but that's dangerously untrue.
Here are 50 ways to prove 1 == 0 in Lean. That compile. Depending on what version you use.
(link in comments) -
Lior Pachter
@lpachter
It's perhaps a minor issue but the population of Greenland is not 30,000.
According to stat.gl/publ/en/BE/202504/contents/Midyear Population.htm, as of 6 months ago it was 56,831. -
Lior Pachter
@lpachter
On a positive note:
www.youtube.com/watch?v=4czjS9h4Fpg -
Lior Pachter
@lpachter
And the culprit is...
-
Lior Pachter reposted
Elliot Glazer
@ElliotGlazer
Why would nlab do this
Mousematician 🌱 Elisha
I'm pretty sure that was the intended interpretation, but I read it as ncatlab.org/nlab/show/2-group first
-
Lior Pachter
@lpachter
If an LLM hallucinates something during a chat session that's AI slop. If a human publishes that hallucination, that's human slop.
The references in this preprint by @LatchBio are therefore human slop.
arxiv.org/html/2512.21907v1
Not a good look. -
Lior Pachter
@lpachter
I've been realizing that I'm becoming increasingly optimistic about AI for scientists because I find AI slop to be much less pernicious than human slop.
Also people are working hard to reduce AI slop while self-interest leads humans to ignore human slop, or even promote it. -
Lior Pachter
@lpachter
"As we grow more dependent on agents to find information and deal with complex negotiations, what will become of our own cognitive skills?"
cyber.harvard.edu/fallsem98/final_papers/Kovacs.html
(from 1998) -
Lior Pachter
@lpachter
This is from a Lex Friedman interview (which the OP should have tagged or at least indicated the source).
And yeah, if I were talking to a person who interviewed like that (i.e. Lex Friedman) I’d stop taking to them. Superficial, subpar interview questions without followups. 👎
Haider.
@haider1
Mathematician Joel David says current AI models are basically zero help for mathematics
They produce garbage answers, then argue they're correct when you point out the exact error
"if i were talking to a person who argued like that, i'd stop talking to them"This media couldn't be displayed. See it on the original post. -
Lior Pachter
@lpachter
I am forthwith an artisanal computational biologist.
Math, Inc.
@mathematics_inc
Terry Tao:
“I do see more and more mass-produced mathematics at scale with that specialization.
I think there will still be traditional, handcrafted mathematics, which will be very valued.
But we will also have this this very complimentary way of doing that” -
Lior Pachter reposted
Daniel Litt
@littmath
I want to explain in down-to-earth terms what this paper is about, since it ultimately boils down to what I think are some really concrete and fundamental questions. 1/n
arXiv math.AG Algebraic Geometry
@mathAGb
Yeuk Hay Joshua Lam, Daniel Litt: Algebraicity and integrality of solutions to differential equations arxiv.org/abs/2501.13175 arxiv.org/pdf/2501.13175 arxiv.org/html/2501.13175
-
Lior Pachter
@lpachter
oof
-
Lior Pachter reposted
Jason Kerwin
@jt_kerwin
This paper appears to be running a diff-in-diff (?) on an RCT with a sample size of 22 (??) and then just reporting all the pre-post differences instead of the treat*post terms (?!?). All the actual results here are null, the rows a/stars in col. 5 are the intervention itself.
Nicholas Fabiano, MD
@NTFabiano
Sleeping 2h earlier significantly improved cognition & mental health.
-
Lior Pachter
@lpachter
Huh. I didn’t realize someone was recording the conversation.
Parmita Mishra 🔬
@parmita
“Show me the mechanism of action.”
“Uh. from the p-p-perturb seq?
right here. Knocks down Gene X, and the cells shift into Cluster 7."
“You used fkn UMAP again. FUCK. Zoom in.”
“Sorry?”
“Zoom. The fuck. In.”
“…Okay.”
“Do you see it?”
“…See what?”
“He doesn’t see it. The cell biology postdoc from Stanford does not know how to see a cell. He stares at a two-million-cell embedding and doesn’t see the conflations he just planted in my S3 bucket.
WHERE is the temporal resolution?”
“The… what?”
“Where is the time? You hit the cell with a perturbation and you measured it once. Once. That’s not a mechanism. That’s a fucking POSTCARD!”
“We sequenced at 72 hours. That’s standard.”
“Standard is not the same as correct. You are aliasing causality. You compressed a dynamic process into a static endpoint and you’re think you will cure metastatic cancer.”
“But the differential expression is significant.”
“WOW! DIFFEWENTIAL EXPRESSION IS SIGNIFICANT! WOW! WHEN THE FUCK IS IT NOT.
Yes, because statistics is Mario Kart.
A fantasy land where causality is optional and variance disappears if you collect enough cells. Real biology has inertia.
Feedback.
Competing pathways.
Do you understand the difference between correlation and mechanism? Or did you flunk out of STAT 101?”
“…I mean, we saw Gene X regulate Pathway Y.”
“No. You saw Pathway Y exist in the same cell after you kicked it down the stairs and waited three days. That’s not even a crime scene, you dimwit. That is a post-mortem autopsy.”
“The model inferred a trajectory--”
“STOP the buffoonery. Do not blame the model.
The model is a mirror.
In this particular case, you can see it mirrors the clusterfuck you just created in my biosafety cabinet.
If the reflection is warped, it’s because your measurement is warped.”
“Pull up the raw counts.”
“…Okay.”
“Scroll. Cell 14,982. Read it to me. What does it say?”
“Uh… mitochondrial genes up, ribosomal genes down-”
“And?”
“…And stress response markers?”
“Yes. Because you poisoned the cell and waited long enough for it to panic.
Where is the early signaling?
Where is the metabolic inflection?
Where is the first irreversible decision?”
“We don’t capture that.”
“Exactly. You built a platform that cannot see the mechanism. YOUR PLATFORM IS FUCKING BLIND.”
“The virtual cell...showed the perturbation effect cleanly.”
“Yes. because your VIRTUAL CELL IS VIRTUAL. it assumes the cell is a bag of transcripts. Do you think metabolites show up? OR AN ISOFORM? AN ISOFORM, THAT WILL DEGRADE IN THE CELL BEFORE YOU EVER REACH YOUR SENSOR? THAT ONE? YOU THINK THAT'S HOW CELLS WORK?”
“So what do you want me to do?”
“Delete the atlas.”
“What?”
“Delete it. The whole thing.”
“But that’s the core result.”
“It’s three weeks of garbage narrative built on a blind instrument. Delete it.”
“…Now what?”
“Now you rebuild the measurement. You observe the cell while the perturbation propagates. You track chemistry, not just transcripts. You capture the first divergence, not the final corpse.”
“That’s… not perturb-seq anymore.”
“Exactly.”
door slams -
Lior Pachter
@lpachter
Erdős was a remarkable mathematician. github.com/teorth/erdosproblems
-
Lior Pachter
@lpachter
Unrelated but Fiji Water is bottled in Fiji in the middle of the Pacific ocean, and then shipped across he world, in this case to Boston. Apparently the "natural artesian water" contains microplastics and BPA. www.newsweek.com/fiji-water-lawsuit-raises-concerns-about…
Mathieu
@miniapeur
Noooooo..
-
Lior Pachter
@lpachter
I highly recommend reading @LedermanHarvey's essay on the meaning of life, the universe, and everything as you ponder the end of 2025, and other things that may soon end as well...
scottaaronson.blog?p=9030 -
Lior Pachter reposted
Grant Slatton
@GrantSlatton
Very fun puzzle, the answer is 2 million
I had codex-cli send the entire Shakespeare corpus through gpt-5.2 in 100 line chunks
"With twenty hundred thousand times more joy" from Romeo and Juliet
Joel David Hamkins
@JDHamkins
Current puzzle in our household: what is the largest number to be explicitly referred to in a Shakespeare play?
-
Lior Pachter
@lpachter
Exhibit 290 in the Epstein files raises some questions about whether he was truly a "source of intellectual exchange and stimulation."
The document is here: www.justice.gov/multimedia/Court Records/Government of th…
Thread on what is in this document below: 1/🧵Post not available here, view on X
Lior Pachter
@lpachter
Epstein's embarrassing advice:
"Algebra is not as important as it used to be. Programming is important."
This has aged about as well as milk.
3/🧵
Lior Pachter
@lpachter
Epstein the liar:
"Unbeknownst to most people today again when I
first started on Wall Street I was a school teacher."
Narrator: Epstein was fired from his job at the Dalton school for poor performance. He also acted inappropriately around girls.
www.nytimes.com/2019/07/12/nyregion/jeffrey-epstein-dalto…
4/🧵Post not available here, view on X
Lior Pachter
@lpachter
I get it. No guilt by association. Just because someone is in a photo with Epstein doesn't mean they knew he pled guilty to soliciting prostitution with a minor. Ok.
But why exactly DID all these "scientists" hang out with him.
Misha Gromov came for this?? Chomsky?? 12/🧵
Lior Pachter
@lpachter
BTW his idea for *the* singular person who could cure cancer was Martin Nowak who has never cured a thing in his life but did promote Epstein x.com/phylogenomics/status/1164549482798653440
Oh, and his reply for how he would train V. Islanders instead of sending them to Harvard? Khan academy.
🧵/🧵.
Jonathan Eisen
@phylogenomics
When, if ever, will @DrMANowak and @harvard do something about their unethical association with rapist Jeffrey Epstein - they took his money, propped up his reputation, and now? crickets - read more from @xeni about this: x.com/xeni/status/1164546818023104512
-
Lior Pachter
@lpachter
I have found ChatGPT5.2 to be incredibly impressive and yet...
-
Lior Pachter reposted
Kanaka Rajan
@KanakaRajanPhD
New paper for #neurips2025!
AI models adjust millions of internal settings to get better at a task. But how are these adjustments determined? For decades, we've mostly used trial & error.
Led by @johnjvastola, we took a different approach...🧵 (1/6)
🔗 openreview.net/forum?id=oMi4uyNOlL -
Lior Pachter reposted
Shriram Krishnamurthi (primary: Bluesky)
@ShriramKMurthi
The best description of @BrownUniversity, from @bessbell. (And to be clear, the heirs and star's kids are invariably also in that carton.)
besskalb.substack.com/p/brown-america -
Lior Pachter
@lpachter
Again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and again and
Lior Pachter
@lpachter
US mass shootings have become so commonplace that sometimes they don't even make front page news. When they do, we learn that "Investigators are still working to determine what led up to the shooting."
You know what led up to the shooting? The distribution of 400 million+ guns. -
Lior Pachter
@lpachter
Would be cool to figure out consciousness but this effort seems woefully underfunded with a way too short timeframe. Oh well.
Astera Institute
@AsteraInstitute
We’re launching Astera Neuro, a new neuroscience research effort led by @doristsao as Chief Scientist. Our aim is to unravel a profound scientific mystery: how the brain transforms sensory inputs into conscious experience.
Advancing this work could illuminate the computational principles that drive perception and cognition and inspire approaches for neuroscience-informed AI research, potentially generating new pathways to AGI.
Astera will support this work with $600M+ over the next decade. Read more: astera.org/neuroscientist-doris-tsao-joins-astera-to-lead… -
Lior Pachter
@lpachter
This guy is an amateur. The pros hallucinate an undergrad who they claim worked on a project in another lab and the PI in the other lab is so out of touch they don't realize that undergrad didn't exist and then the pros get authorship on the project in the other lab.
Retraction Watch
@RetractionWatch
How to juice your Google Scholar h-index, preprint by preprint.
retractionwatch.com/2025/12/08/how-to-juice-your-google-s… -
Lior Pachter
@lpachter
Umm... this was definitely not my experience. liorpachter.wordpress.com/2018/01/17/terry-speed-a-male-f…
Sri Kosuri
@srikosuri
Will probably be flamed for this, but I agree with this take. Being an assistant professor was the easiest job I’d ever had in terms of number of hours and levels of stress (financial and otherwise), especially compared to postdoc or startup.
-
Lior Pachter reposted
Dr. Jean Fan
@JEFworks
Ok so I made this list of disliked academic flexes into a song to the tune of Bad Guy by Billie Eilish.
It's...Bad PI: www.youtube.com/shorts/W-A4nmZ4guQ
🤣🤣🤣
Lior Pachter
@lpachter
Academic flexes that I dislike: 1/🧵
-
Lior Pachter reposted
Oliver Steinbock
@SteinbockGroup
I'm curious ... do you think there is a number of papers/year above which it is unlikely that the author met common authorship criteria, including substantial intellectual contribution, participation in writing and critically revising, ...? Where would you draw that line?
-
Lior Pachter reposted
Jonathan Eisen
@phylogenomics
The Tree of Life: A ton to be thankful for phylogenomics.blogspot.com/2025/11/a-ton-to-be-thankful-f…
-
Lior Pachter reposted
Faisal AlZaben • فيصل الزبن
@polymorfaisal
It's curious to see how slowly biology pubs are adopting modern AI methods. Journals still tend to reward work that's essentially a small demo repo or a straightforward application of an existing tool.
I think we're overdue for a higher bar.
With all due respect. -
Lior Pachter reposted
Faisal AlZaben • فيصل الزبن
@polymorfaisal
It gets worse 😳 I get review cycles are slow but the newest model is over a year old.
LLM configs is out of date AND includes deprecated models...
"we built an LLM-provider-agnostic Python package that can use any available LLM by changing just one line of code". Great
Lior Pachter
@lpachter
"We designed an LLM agent that attempts to determine cluster resolution automatically from UMAP plots."
Like seriously. Seriously. -
Lior Pachter
@lpachter
My student Catherine Felce will be defending her thesis @caltech.edu next week. If you're in the area consider attending; it will be a treat!
Cat's recent work:
journals.aps.org/pre/abstract/10.1103/r8bd-lhm3
www.biorxiv.org/content/10.1101/2025.11.24.690267v1
www.biorxiv.org/content/10.1101/2025.11.14.688548v1 -
Lior Pachter reposted
Mark Borris Aldonza (@precursorcell.bsky.social🦋)
@PrecursorCell
I guess sometimes, you have to let an LLM look at a UMAP plot just to confirm it shouldn’t
-
Lior Pachter
@lpachter
Multi-Modal (single-cell genomics)... flip it around... Wicked Witch! 😈
-
Lior Pachter reposted
David Garfield (@dagarfield@mstdn.social and post)
@dagarfield
My LLM reiterates on UMAP until it looks like a flower.
-
Lior Pachter
@lpachter
"We designed an LLM agent that attempts to determine cluster resolution automatically from UMAP plots."
Like seriously. Seriously. -
Lior Pachter reposted
(((ل()(ل() 'yoav))))👾
@yoavgo
a key lesson from this is that looking at the data should be the first thing you do, not a last resort after you try to debug some surprising low scores. it really amazes me how many people neglect to do this very obvious thing, and how unintuitive this advice is to them.
Lei Yang
@diyerxx
Got burned by an Apple ICLR paper — it was withdrawn after my Public Comment.
So here’s what happened. Earlier this month, a colleague shared an Apple paper on arXiv with me — it was also under review for ICLR 2026.
The benchmark they proposed was perfectly aligned with a project we’re working on.
I got excited after reading it. I immediately stopped my current tasks and started adapting our model to their benchmark.
Pulled a whole weekend crunch session to finish the integration… only to find our model scoring absurdly low.
I was really frustrated. I spent days debugging, checking everything — maybe I used it wrong, maybe there was a hidden bug.
During this process, I actually found a critical bug in their official code:
* When querying the VLM, it only passed in the image path string, not the image content itself.
The most ridiculous part? After I fixed their bug, the model's scores got even lower!
The results were so counterintuitive that I felt forced to do deeper validation. After multiple checks, the conclusion held: fixing the bug actually made the scores worse.
At this point I decided to manually inspect the data. I sampled the first 20 questions our model got wrong, and I was shocked:
* 6 out of 20 had clear GT errors.
* The pattern suggested the “ground truth” was model-generated with extremely poor quality control, leading to tons of hallucinations.
* Based on this quick sample, the GT error rate could be as high as 30%.
I reported the data quality issue in a GitHub issue. After 6 days, the authors replied briefly and then immediately closed the issue.
That annoyed me — I’d already wasted a ton of time, and I didn’t want others in the community to fall into the same trap — so I pushed back. Only then did they reopen the GitHub issue.
Then I went back and checked the examples displayed in the paper itself.
Even there, I found at least three clear GT errors.
It’s hard to believe the authors were unaware of how bad the dataset quality was, especially when the paper claims all samples were reviewed by annotators. Yet even the examples printed in the paper contain blatant hallucinations and mistakes.
When the ICLR reviews came out, I checked the five reviews for this paper.
Not a single reviewer noticed the GT quality issues or the hallucinations in the paper's examples.
So I started preparing a more detailed GT error analysis and wrote a Public Comment on OpenReview to inform the reviewers and the community about the data quality problems.
The next day — the authors withdrew the paper and took down the GitHub repo.
Fortunately, ICLR is an open conference with Public Comment. If this had been a closed-review venue, this kind of shoddy work would have been much harder to expose.
So here’s a small call to the community:
For any paper involving model-assisted dataset construction, reviewers should spend a few minutes checking a few samples manually. We need to prevent irresponsible work from slipping through and misleading everyone.
Looking back, I should have suspected the dataset earlier based on two red flags:
* The paper’s experiments claimed that GPT-5 has been surpassed by a bunch of small open-source models.
* The original code, with a ridiculous bug, produced higher scores than the bug-fixed version.
But because it was a paper from Big Tech, I subconsciously trusted the integrity and quality, which prevented me from spotting the problem sooner.
This whole experience drained a lot of my time, energy, and emotion — especially because accusing others of bad data requires extra caution.
I’m sharing this in hopes that the ML community remains vigilant and pushes back against this kind of sloppy, low-quality, and irresponsible behavior before it misleads people and wastes collective effort.
#iclr #iclr2026 #neurips #cvpr #openreview #machinelearning #llm #vlm -
Lior Pachter
@lpachter
Wrong. Completely wrong. And now dangerously wrong.
Lior Pachter
@lpachter
Turns out this piece by @wespegden, @sdbaral and @VPrasadMDMPH was wrong. Completely wrong. blogs.bmj.com/bmj/2021/05/07/covid-vaccines-for-children-…
-
Lior Pachter
@lpachter
I have no skin in this Berkeley-Stanford game but always perform a control.
I guess Stanford students get trained by Berkeley students who were trained by Stanford students, etc.
Actually, the first CS department in the US was at Purdue University (1962).
Jelani Nelson
@minilek
Stanford is a fantastic place to study CS -- you get to be trained by Berkeley students!
-
Lior Pachter reposted
Dr. Jean Fan
@JEFworks
I love this list of favorite academic flexes... so I made it into a song to the tune of My Favorite Things from The Sound of Music 😂
My Favorite Things (Academic Edition) 🎶: youtube.com/shorts/v9qQvznDuTY
(Is there a good tune to go with the list of things we dislike? Let me know!) -
Lior Pachter
@lpachter
Omg this is brilliant.
Dr. Jean Fan
@JEFworks
I love this list of favorite academic flexes... so I made it into a song to the tune of My Favorite Things from The Sound of Music 😂
My Favorite Things (Academic Edition) 🎶: youtube.com/shorts/v9qQvznDuTY
(Is there a good tune to go with the list of things we dislike? Let me know!) -
Lior Pachter reposted
Asaf Levy אסף לוי اسف ليڤي
@AsafLevyHUJI
Professors whose slides have a single panel with a clear title and not the 7 small unreadable panels from their paper.
-
Lior Pachter
@lpachter
Ok, here it is: 1/🧵
Dvir Aran
@dvir_a
Great thread. Now do the things you do like :)
Post not available here, view on X
Lior Pachter
@lpachter
Github repositories that have the code for a paper in a form that it can easily be run (ideally on Google Colab or similar) and with data accessible and metadata properly annotated.
Lior Pachter
@lpachter
Sepakers who end on time.
Post not available here, view on X
Lior Pachter
@lpachter
Scientists who cite all the papers relevant to the work their work and nothing else.
Lior Pachter
@lpachter
Collaborations where the whole is greater than the sum of the parts.
-
Lior Pachter
@lpachter
Academic flexes that I dislike: 1/🧵
Lior Pachter
@lpachter
Starting a talk by listing dozens of affiliations and consulting gigs under the guise of declaring COIs. Absolutely declare COIs... a laundry list of things you've done over the past 30 years is not it.
Lior Pachter
@lpachter
Posting photos with colleagues taken at conferences in exotic locales. Showing off how you wasted taxpayer grant money so that colleagues think you're accomplished by association is not the flex you think it is.
Post not available here, view on X
Lior Pachter
@lpachter
Going on tons of podcasts with dubious characters under the guise of "sharing science with the public" but then only exaggerating about your own work is poor form and counterproductive (for the public and for you).
Lior Pachter
@lpachter
-
Lior Pachter reposted
sina
@sinabooeshaghi
Biology by Biorender, say hello to Biology by Banana.
Derya Unutmaz, MD
@DeryaTR_
Wow! NanoBanana Pro is incredible! This is a perfect cartoon depicting the activation of a T cell with the full signaling cascade! I only asked it to draw the interaction & show the signaling & it figured out the rest!
I am declaring that image-level AGI has now been achieved! -
Lior Pachter reposted
Anshul Kundaje
@anshulkundaje
"* The header "Vaccines do not cause autism" has not been removed due to an agreement with the chair of the U.S. Senate Health, Education, Labor, and Pensions Committee that it would remain on the CDC website."
Unbelievable jokers causing immense damage. -
Lior Pachter
@lpachter
Astonishing update to the CDC page on vaccines and autism. In demanding a p-value to *accept* the null hypothesis, harm will be done via promotion of vaccine hesitation.
A tragedy unfolding.
www.cdc.gov/vaccine-safety/about/autism.htm -
Lior Pachter reposted
Andrew Gelman et al.
@StatModeling
This guy’s mad about fake research, and he should be. Research incompetence, research fraud, and the promotion of fraudulent or incompetent work . . . these are not victimless crimes.
statmodeling.stat.columbia.edu/2025/11/19/this-guys-mad-a… -
Lior Pachter
@lpachter
I had a research breakthrough recently and was able to glimpse something that very few people have ever seen: the virtual cell.
-
Lior Pachter
@lpachter
Thinking about math skills among students in the past vs. today.
When I was in high school a kid in my math class used his calculator to multiply a number by 1.
This year a 17-year old kid (@Hannahcairomath) disproved Mizohata–Takeuchi. -
Lior Pachter reposted
Nature Methods
@naturemethods
Monod fits biophysically motivated models to single-cell transcriptomics data, providing insights into gene expression dynamics. @goringennady @lpachter @mariacarilli @johnjvastola
www.nature.com/articles/s41592-025-02832-x -
Lior Pachter
@lpachter
Watson's passing seems to have brought misogynists out of the woodwork to criticize Rosalind Franklin's math ability. Not only did she understand Fourier analysis & its connection to X-ray crystallography, she wrote a single-author paper about it in 1949: journals.iucr.org/paper?a00229
-
Lior Pachter
@lpachter
One of the curiosities of genomics history is that James Watson was vehemently against cDNA sequencing (for ESTs), and fought with Craig Venter & Bernadine Healy who championed it. Tl;dr Watson ended up resigning from the HGP, Venter plowed ahead... and we now have #scrnaseq.
Post not available here, view on X
Lior Pachter
@lpachter
Worth reading Baltimore's report at the time: www.genome.gov/sites/default/files/archiespace2024-01/ADH…
"Consideration should be given to... cDNA maps. The Committee recognizes the many challenges... it is not yet known to what extent eukaryotic cells rely on control at the transcriptional level."
Lior Pachter
@lpachter
Report from 1992 in Nature on the cDNA controversy: www.nature.com/articles/356549a0.pdf
-
Lior Pachter reposted
Tim Stearns
@StearnsLab
Thread: a letter from Francis Crick to Jim Watson about a draft of Watson’s book. profiles.nlm.nih.gov/101584582X137
Tim Stearns
@StearnsLab
In the interest of promoting an understanding of the history, this scathing letter from Crick to Watson in 1967 is a remarkable document. Crick lays out a detailed criticism of the draft of Watson's "Double Helix" (provisionally titled "Honest Jim" here). profiles.nlm.nih.gov/ps/access/SCBBKN.pdf
-
Lior Pachter
@lpachter
Quite something to see racists and misogynists who like to scream "merit based hiring only!" turn around to idolize Watson who rejected it. His approach to hiring scientists was “whenever you interview fat people, you feel bad, because you know you’re not going to hire them.”
Lior Pachter
@lpachter
Scientific breakthroughs are rarely unique; someone else would’ve made them soon enough. But when prominent scientists cause harm, that harm isn’t inevitable; the world might simply have been better had the harm not been inflicted.
Lior Pachter
@lpachter
I arrived in Berkeley in 1999, and he gave his talk in the MCB department just a year later. The seminar is recounted here: www.sfgate.com/science/article/Nobel-Winner-s-Theories-Ra…
Lior Pachter
@lpachter
Maybe Watson on merit based hiring is like Watson's sexism and racism. He just said he was against merit based hiring but one was supposed to believe that he was actually for merit based hiring.
-
Lior Pachter
@lpachter
-
Lior Pachter
@lpachter
I wrote a review of James Watson's obituary by @CSHL: liorpachter.wordpress.com/2025/11/08/review-of-cold-sprin…
-
Lior Pachter
@lpachter
Scientific breakthroughs are rarely unique; someone else would’ve made them soon enough. But when prominent scientists cause harm, that harm isn’t inevitable; the world might simply have been better had the harm not been inflicted.
-
Lior Pachter
@lpachter
“A problem worthy of attack, proves its worth by fighting back!” – Paul Erdos arxiv.org/abs/2511.02864
And indeed the Erdos-Szekeres happy ending problem is still fighting. Having said that, this is an astounding paper. -
Lior Pachter reposted
Michael 英泉 Eisen
@mbeisen
I’m saying that ad-hoc supervision by volunteers who hold the guest speakers in high esteem and who clearly don’t believe this supervision is necessary is not something a hosting institution can rely on. And I say this out of experience not theory.
-
Lior Pachter reposted
Michael 英泉 Eisen
@mbeisen
I'm a big fan of Berkeley Math Circle - one of my kids did it for a while - but they also did gymnastics, and let me tell you, you want people working with your kids to go through rigorous background checks.
Doris Tsao
@doristsao
Unbelievable: the famed Berkeley Math Circle is being forced to shut down due to a bureaucratic requirement where a guest lecturer giving an hour long lesson needs to be officially fingerprinted. How is fingerprinting even still a thing in the 21st century?
Chancellor Lyons @richlyons: can you see the absurdity of the situation and figure out a solution?
www.dailycal.org/news/campus/genius-producing-math-progra… -
Lior Pachter
@lpachter
Serge Lang proselytized that HIV doesn’t cause AIDS. So I certainly don’t wish to see him cloned into an LLM.
Amin Karbasi
@aminkarbasi
I heard plenty of similar stories about him during my time at Yale. At this rate, we might end up replacing Serge Lang with our favorite LLM. The bigger shift, though, is access: what used to be reserved for a small circle of people who knew the “Langs” of the world will soon be available to everyone.
-
Lior Pachter
@lpachter
In the debate of whether AI is more or less energy efficient than humans it’s worth considering how much energy it would take to raise a human capable of accelerating @wtgowers.
Timothy Gowers @wtgowers
@wtgowers
I crossed an interesting threshold yesterday, which I think many other mathematicians have been crossing recently as well. In the middle of trying to prove a result, I identified a statement that looked true and that would, if true, be useful to me. 1/3
-
Lior Pachter
@lpachter
Wishing everyone a happy Halloween 🎃
-
Lior Pachter reposted
Lance Fortnow
@fortnow
A paper by Babai on the topic from 1990. Section 16 is particularly on point. people.cs.uchicago.edu/~laci/papers/email90.pdf
-
Lior Pachter reposted
Kalle Leppälä
@kalle_leppala
In principle I support swindling money from idiot tech bros, but the eugenics company doing embryo selection for IQ is bad. It's bad that they achieve their outlandish predictive power by statistical trickery with assumed latent factors. It's bad that they control for...
-
Lior Pachter reposted
Anshul Kundaje
@anshulkundaje
Calling on all scientists to call out hype, grift & BS when u see it. If u don't, u risk anti-science grifters leveraging overhyped & underdelivered promises & shoddy science to destroy public trust in the enterprise. Keep science & science comm grounded & truthful.
-
Lior Pachter
@lpachter
tl:dr this is an excerpt from a physics textbook.
Anthony Bonato
@Anthony_Bonato
"Dad ‘Embarrassed’ To Be Struggling With Daughter’s Third-Grade Math Problem"
-
Lior Pachter reposted
Talia Ringer 🕊🪬
@TaliaRinger
This performance art piece that calls itself a conference must be a gold mine for hallucinated citations. First paper I looked at, for example, and first citation I checked (citation 14). Just totally made up authors
James Zou
@james_y_zou
#agents4science received over 300 submissions and accepted 48 papers. Check them out here: openreview.net/group?id=Agents4Science/2025/Conference
Many of these AI agent-written papers are quite interesting! The online conference will be on Oct 22; you can register for free at stanford.zoom.us/webinar/register/WN_QWJZ4DhvRHCHfR4rFL24PQ -
Lior Pachter
@lpachter
I listened to the Soundcloud and FWIW I think Pythagoras would have preferred Bach.
-
Lior Pachter
@lpachter
The @HCSSiM program has been an educational gem for decades, leading "to a world where people value the truth and the project of finding the truth". www.youtube.com/watch?v=lnL26yZGX1g
-
Lior Pachter reposted
Mel Andrews
@bayesianboy
My article on AI for science, in which I characterize a deviant notion of scientific objectivity rooted in the impossible ideal of theory-free inference, is available now open-access in Erkenntnis link.springer.com/article/10.1007/s10670-025-01010-x
-
Lior Pachter
@lpachter
Strange that the MIT AI BIO Frontiers Forum has no women in their lineup, especially considering that many of the top people in the field right now are women. 👀
-
Lior Pachter
@lpachter
I recently visited the town of Zbąszyń in Poland. I took some time there to reflect on the plight of refugees and to bear witness for the dead and the living. This led me to write something up about its history.
liorpachter.wordpress.com/2025/09/22/zbaszyn/ -
Lior Pachter
@lpachter
24
Lior Pachter
@lpachter
Thinking of my graduate school classmate Danny Lewin today. He was likely the first victim of the September 11, 2001 attacks, just about this time 22 years ago. en.wikipedia.org/wiki/Daniel_Lewin
-
Lior Pachter reposted
SciencesNetwork
@SciencesNetwork
Reference-based variant detection with varseek
www.biorxiv.org/content/10.1101/2025.09.03.674039 -
Lior Pachter
@lpachter
In a new work with @Josephmrich and Conrad Oakes we tackle the problem of how to best organize alluvial plots. We formalize two optimization problems and develop a solution for them based on the neighbornet algorithm, implemented in the program wompwomp: github.com/pachterlab/wompwomp
Post not available here, view on X
Lior Pachter
@lpachter
The underlying optimization problems are NP-hard, but for many visualization tasks can probably be solved optimally. It was fun returning to Bryant and @HusonDaniel's NeighborNet as a good heuristic, but there is certainly room for improvement.
Lior Pachter
@lpachter
Brilliant work by @Josephmrich & Conrad Oakes working out the math and implementing wompwomp. It has been submitted to @Bioconductor and in the future we may also implement it in Python to facilitate usage with single-cell genomics packages. Any feedback will be appreciated. Thx!
-
Lior Pachter
@lpachter
I got notified of a spike of viewers on this post from 11 years ago liorpachter.wordpress.com/2014/04/17/does-researching-cas…
Indeed the study I critiqued didn't replicate www.jneurosci.org/content/35/4/1505
"The press may not cite studies that do not find sensational effects, but these studies are still extremely important" -
Lior Pachter
@lpachter
Same manipulation that Ferrero uses to advertise Tic Tacs as sugar free (and have people believe them).
Sasha Gusev
@SashaGusevPosts
I wrote about how genetic risk works in the context of embryo selection and how people often think about it all wrong. A short 🧵:
-
Lior Pachter
@lpachter
If this is how the math on the Manhattan project was done they'd probably still have made the movie but Oppenheimer would have been played by Rowan Atkinson.
-
Lior Pachter
@lpachter
Here we go again. Another paper showing that putting arrows on UMAPs (RNA velocity) is biological gobbledygook. www.biorxiv.org/content/10.1101/2025.08.02.668272v1
How many more such papers are needed? -
Lior Pachter
@lpachter
"Every human life is equally valid" says person whose company ranks embryos from "most valid" to "least valid" according to scores (whose computation the person claims is a trade secret), so that the "least valid" embryos can be discarded.
-
Lior Pachter reposted
sina
@sinabooeshaghi
A company claiming GPT-5 is a “legitimate PhD-level expert” has clearly never sent it to group meeting to defend this barplot.
52.8% > 69.1% == 30.8% -
Lior Pachter
@lpachter
This is not true. Identical twins do not share 100% identical DNA. Close but no dice. www.nature.com/articles/s41588-020-00755-1
Toan Truong
@LearningToan
The science is simple but powerful:
Identical twins share 100% of their genes.
Fraternal twins share 50% (like regular siblings).
By comparing how similar identical twins are versus fraternal twins, researchers can measure genetic influence on any trait. -
Lior Pachter reposted
Graham Coop
@Graham_Coop
These embryo-selection startups are clearly feeding into an alt-right ecosystem that revels in techno-futurism much as such movements have in the past. GWAS participants & parents navigating IVF deserve better than being used as tools to attract the attention of edge lords.
-
Lior Pachter reposted
Graham Coop
@Graham_Coop
It is depressing, but all too predictable, how swiftly we’ve gone from the Social Science Genetic Association Consortium offering reassurances about the uses of behavioural polygenic scores to one of their lead authors marketing embryo selection for IQ
-
Lior Pachter reposted
aru
@arubikscube
whoops, almost found SRA IDs manually
god bless easy to use command line tools
Lior Pachter
@lpachter
(1/9) `ffq` (Fetch FastQ) is a command line tool for finding #sequencing data from public databases like SRA / GEO / @ENASequence / @NCBI / @DDBJ_topics / @ENCODE_NIH. In this quote-tweet thread, @agalvezm, @sinabooeshaghi and I describe the new features.
github.com/pachterlab/ffq -
Lior Pachter
@lpachter
זכרונו לברכה
Lior Pachter
@lpachter
Mind boggling to consider that Tom Lehrer, who is 94 years old, retired from writing and performing music... 50 years ago (!)
He has placed all of his songs in the public domain on this website: tomlehrersongs.com -
Lior Pachter
@lpachter
This is a good article and I agree with its argument for comprehensible research.
Richard McElreath 🐈⬛
@rlmcelreath
How can we reform science? I have some ideas. But I am not sure you’ll like them, because they don’t promise much. elevanth.org/blog/2025/07/09/which-kind-of-science-reform/
-
Lior Pachter reposted
SternLab @sternadi.bsky.social
@SternLab
Given incredibly difficult, conflict-filled years,it makes me proud to present @BZman101 beautiful work on intra-host evolution in SARS-CoV-2. Briefly,we find evolution differs among patients: in some nothing happens; in others rampant adaptive evolution. academic.oup.com/ve/article/doi/10.1093/ve/veaf047/8162928
-
Lior Pachter
@lpachter
I recently learned of the Damm algorithm which is very cool. en.wikipedia.org/wiki/Damm_algorithm
-
Lior Pachter reposted
Kenny Workman
@kenbwork
Sina's recorded defense is an awesome resource for those interested: www.youtube.com/watch?v=xD-lZGE1fEg. He came into the space with a MechE background and questioned engineering assumptions + vendor prices with single cell tooling
-
Lior Pachter reposted
Gennady Gorin
@GorinGennady
LR-Kallisto spotted!
Stephen Turner
@strnr
Dogme: A nextflow pipeline for reprocessing nanopore RNA and DNA modifications pubmed.ncbi.nlm.nih.gov/40501581/ 🧬🖥️🧪 github.com/mortazavilab/dogme
-
Lior Pachter
@lpachter
aka PCA
Peyman Milanfar
@docmilanfar
Most tech people know about ordinary least-squares (OLS), but not so much about *total* least-squares (TLS)
They measure fit differently: OLS minimizes sum of sq. distances along dependent variable; whereas TLS minimizes sum of orthogonal distances from data to fit line
1/2 -
Lior Pachter
@lpachter
The confiscation of cell phones from children at schools & summer camps is largely the result of a national campaign by @JonHaidt.
But cell phones are safety devices used for notification & early warnings of active shooters, earthquakes, floods, tornados, tsunamis, wildfires, .. -
Lior Pachter
@lpachter
This week in academia, a not so short🧵...
1. Staff reductions and other cost cutting measures coming to Brown University www.highereddive.com/news/brown-university-cost-cutting-f…Post not available here, view on X
Lior Pachter
@lpachter
6. Layoffs looming at Stanford University www.sfgate.com/bayarea/article/stanford-budget-cuts-layof…
Lior Pachter
@lpachter
7. Harvard University announces more layoffs www.thecollegefix.com/harvard-announces-more-layoffs-as-f…
Post not available here, view on X
Lior Pachter
@lpachter
16. UCSF to lay off 200 employees www.kron4.com/news/bay-area/ucsf-to-lay-off-about-200-emp…
Lior Pachter
@lpachter
17. University of Washington warns of potential layoffs www.kuow.org/stories/uw-president-warns-of-budget-crisis-…
-
Lior Pachter
@lpachter
Very, very sad to hear of the passing of David C. Kelly (known as Kelly). A real mensch. Here is a beautiful remembrance by my former math professor @JimPropp: mathenchant.wordpress.com/2025/06/17/remembering-kelly/
Lior Pachter
@lpachter
I cherish these two iconic $2 notes I received from Kelly a few years ago.
-
Lior Pachter reposted
Rasmus Nielsen
@ras_nielsen
Got several responses exploring which software can deal with this. My point was that students should not rely on the precision of any standard software/functions but should learn how to maintain precision and avoid underflow/overflow before starting to code.
Rasmus Nielsen
@ras_nielsen
Quantitative graduate students should start by studying this. Would save them endless frustrations.
-
Lior Pachter
@lpachter
I wrote a review of a recent paper on false discovery and multiple testing correction. liorpachter.wordpress.com/2025/06/16/reply-to-reply-to-fa…
-
Lior Pachter
@lpachter
Call for nominations for the Trevisan prize: cs.unibocconi.eu/call-nominations-trevisan-prize-2025
-
Lior Pachter reposted
Rasmus Nielsen
@ras_nielsen
Preliminary program now available for ProbGen 2026: probgen2026.github.io. We have an amazing set of speakers let by our keynotes Sally Otto and @jkpritch. 1/8
-
Lior Pachter
@lpachter
As @emukamel and Yu point out, #methodsmatter. They are spot on here: www.nature.com/articles/s41586-025-08988-y
-
Lior Pachter reposted
Laura Luebbert, PhD
@NeuroLuebbert
🚨We’re hiring 🚨a Bioinformatics Software Engineer to scale up viral detection pipelines on DNA- and RNA-seq data!
The role involves machine learning, lots of cloud computing, and contributions to open-source tools.
If this sounds fun or you know someone, please reach out! -
Lior Pachter reposted
-
Lior Pachter
@lpachter
In hindsight, this was my most underrated tweet.
Lior Pachter
@lpachter
I'm saying that his "all papers are wrong" paper is wrong and so is his most recent paper which is frustrating because as you point out he is trying to make his own point but then he is failing at that because in fact not all papers are wrong, only his seem to be.
-
Lior Pachter
@lpachter
The news yesterday was a horrible gut punch. Devastated to learn of @sayanmukher's passing. Every conversation with him was a delight; he was brilliant and deep and kind. A huge loss. Heartbreaking. www.mis.mpg.de/news/loss-sayan-mukherjee
-
Lior Pachter
@lpachter
Apropos of nothing: less than 10% of pilots are women (5% for airline). And yet it's male pilots (specifically older & w/ experience) who are more likely to be involved in a fatal accident. www.sciencedirect.com/science/article/pii/S0001457510003672
There's a pilot shortage. Shouldn't one try hard to recruit women? -
Lior Pachter
@lpachter
If you need to sample reads (with or without replacement) from FASTQ files you may find the fastQpick utility by @Josephmrich super useful: github.com/pachterlab/fastQpick
-
Lior Pachter
@lpachter
The Pasadena synagogue that was destroyed in the Eaton fire was in.. Altadena. That's bc Pasadena had exclusionary & antisemitic convenants in the 20th century, e.g. "premises shall not [be].. occupied or used by any person or persons other than those of the Caucasian race".
-
Lior Pachter reposted
Caltech
@Caltech
REVISED Water Advisory—Pasadena City and Pasadena Water and Power Officials have confirmed this morning that the tap water on the Caltech campus is safe to drink and cook with; it has not been impacted by the Eaton Fire. For more information, visit caltech.edu/emergency.
-
Lior Pachter
@lpachter
A map of where it is safe (January 9th 12pm) to currently drink water in Pasadena and Altadena.
White: safe
Grey: do not drink water (not even if treated yourself by boiling etc.)
Source: www.cityofpasadena.net/city-manager/news/pasadena-drinkin… -
Lior Pachter reposted
Caltech
@Caltech
Due to the Eaton Fire in Altadena and strong winds in the area, Caltech’s campus in Pasadena will be closed for all nonessential operations on Wednesday, January 8. For more information, visit Caltech's Emergency Updates page.
www.caltech.edu/emergency -
Lior Pachter
@lpachter
My guess is that this will also end up being retracted.
Colin Wright
@SwipeWright
If you're unverified and want to troll me or reply, it will have to be via quote-post!
-
Lior Pachter
@lpachter
TIL that Jimmy Carter was the only president to have taken graduate-level coursework in science / engineering.
-
Lior Pachter reposted
Hattie Chung
@hattaca
Read the preprint here: www.biorxiv.org/content/10.1101/2024.12.15.628550v1
-
Lior Pachter reposted
Cell Press
@CellPressNews
Length biases in single-cell RNA sequencing of pre-mRNA. Check out this research by @lpachter & @GorinGennady in @BiophysReports #cellbio2024 www.cell.com/biophysreports/fulltext/S2667-0747(22)00054-…
-
Lior Pachter
@lpachter
The banality of racism. CEO of Orchid "..it’s super important. The pop. of all of the places we love is shrinking. In 50 years, 30 years, you’ll have half as many people in places that you love. Society will collapse"
Hello? Africa? We don't love? www.wired.com/story/this-woman-will-decide-which-babies-a…Post not available here, view on X
Lior Pachter
@lpachter
I wrote about Orchid 3+ years ago. Hard to believe they are still around... or maybe not hard to believe.
Lior Pachter
@lpachter
The amoral nonsense of Orchid’s embryo selection liorpachter.wordpress.com/2021/04/12/the-amoral-nonsense-…
Lior Pachter
@lpachter
Their website www.orchidhealth.com/careers says "The technology behind Orchid has been developed by experts in medical genetics, embryology, and assisted reproductive technology."
Then they list "expert" Jan Liphardt who is an expert on.. Bitcoin and DeFi. 👀 janliphardt.com/author/jtliphardt/ -
Lior Pachter reposted
Artem Chernikov
@archernikov
There will be a Very Informal Gathering of Logicians (VIG 2025) at UCLA, Fri, Feb 7 to Sun, Feb 9, 2025. The 22nd in a series of biennial logic meetings at UCLA, this event honors Alexander Kechris on the occasion of his formal retirement from Caltech.
www.math.ucla.edu/~ineeman/Conf/VIG2025/ -
Lior Pachter
@lpachter
Another example of the different results one gets from a Scanpy vs. Seurat analysis, in this case a Visium HD dataset from academic.oup.com/bioinformatics/advance-article/doi/10.10…
13 vs. 20 clusters. And qualitatively different results, corroborating what we saw in www.biorxiv.org/content/10.1101/2024.04.04.588111v1 -
Lior Pachter
@lpachter
Cool slide.
Saket Choudhary
@saketkc
We also of course cover single-cell RNA-seq and its nitty-gritty! docs.google.com/presentation/d/1BZfI5fvb1n_GNwTOqcTFtollp… 7/n
-
Lior Pachter reposted
Saket Choudhary
@saketkc
Had a lot of fun teaching DH 607 Introduction to computational multi-omics for the first time @KCDH_IITB/@iitbombay!
All course material (slides/videos) is available on the website: saket-choudhary.me/DH607/syllabus
Some highlights/thoughts [1/n] -
Lior Pachter reposted
Gennady Gorin
@GorinGennady
Yes. It would be nice to have something like this, given that this is a standard part of every single-cell workflow. Even a coarse implementation would be a good starting point. But instead we have a hundred variants of UMAP and a few scores of nonquantitative foundation models.
-
Lior Pachter reposted
Llan Fair
@FairLlan
@AdrianoAguzzi @lpachter just following up. The inaccessibility of this library is limiting science that might otherwise be done. Researchers should not have to write a proposal to you, and/or pay you/travel to your lab to access a very useful plasmid library.
-
Lior Pachter reposted
Prashant Bhat
@_prashantbhat
Hi! I’ll be speaking at @harvardmed Department of Biomedical Informatics on December 5th about genome organization and splicing/speckles. Thank you to the organizers for the invite! Harvard/MIT/affiliate trainees can RSVP to attend with this link:
forms.gle/LopwGCc8RbuCqSu37 -
Lior Pachter
@lpachter
This is messed up.
Junyue
@junyue_cao
🚨 The account @junyue__cao is a fake impersonation of me (@junyue_cao), with two underscores added between my first and last name. It is using my photos and pretending to represent my lab. Please unfollow this account and disregard any messages it sends. Thank you! 🚨
-
Lior Pachter
@lpachter
Interesting, in the year 2024, to read the account of Dorothy Thompson's "Who Goes Nazi?" game from 1941: harpers.org/archive/1941/08/who-goes-nazi/
Of course it's fun sport to go through individuals A. through L., but she forgets to play the game with one of the key people in the room... herself. -
Lior Pachter reposted
Dr. 🅰️aron Thomason
@DrAaronThomason
I provided R scripts on @RetractionWatch proving Poore used two different data sets, one predictive with batch effects and 58 no information genomes, and another without these that was used tlfor the PCA plot in the paper. Unbelievable error.
-
Lior Pachter
@lpachter
The moral arc of the universe is long, but it bends towards justice non-monotonically.
-
Lior Pachter reposted
Roland Dunbrack 🏳️🌈 @rolanddunbrack.bsky.social
@RolandDunbrack
And AF3 code is finally released. www.science.org/content/article/google-deepmind-releases-…
-
Lior Pachter
@lpachter
I have similarly had a single research success in mathematics, with GPT (o1) improving a bound.
Daniel Litt
@littmath
I have had one honest success using GPT (o1) to do research mathematics—I asked it for a counterexample to a strengthening of a conjecture I was thinking about, and it gave me a correct(!) counterexample.
-
Lior Pachter reposted
Yad Vashem
@yadvashem
#kristallnacht
At 9:30 AM #otd 10 November 1938, some fifty rioters arrived at the Jewish orphanage in Dinslaken and started vandalizing and destroying the place
There were 32 children and 15 members of staff at the orphanage that day
Learn more: www.yadvashem.org/yv/en/exhibitions/kristallnacht/jewish-… -
Lior Pachter reposted
Valentine Svensson
@vallens
After some outage due to hosting moves, the single cell studies database Google Sheet frontend is back: nxn.se/single-cell-studies/gui
-
Lior Pachter reposted
Jack
@leary_jack
very interested in this, just wish it was open-source :/
Dylan Webster
@DylanPWebster
Super excited about this collab! Try it for free on cloud.10xgenomics.com/cloud-analysis and look out for more in the next Cell Ranger release. (Let us know what you think!)
Thanks to everyone at @10xGenomics and @broadinstitute who made this happen!
#singlecell #bioinformatics #genomics -
Lior Pachter
@lpachter
The Göttingen synagogue was burned down during Kristallnacht 86 years ago. In its place there is now a sculpture that beautifully evokes the Star of David (blue lines added by me).
Lior Pachter
@lpachter
This past April I visited Germany to go to a conference in Bonn, and to present several lectures at two Max Planck Institutes. In between I visited several sites of synagogues burned down on Kristallnacht, 85 years ago today. These "Ehemalige Synagoge" are all over Germany. 1/
-
Lior Pachter
@lpachter
People frequently tell me that they agree with my critiques of UMAP but that there really isn't much they can do, because reviewers demand them. But do reviewers demand them?
Here is a review from the CellOT paper (static-content.springer.com/esm/art%3A10.1038%2Fs41592-02…) with the opposite ask. -
Lior Pachter reposted
Andrew Gelman et al.
@StatModeling
Fake data on the honeybee waggle dance, followed by the inevitable “It is important to note that the conclusions of our studies remain firm and sound.”
statmodeling.stat.columbia.edu/2024/11/07/fake-data-on-th… -
Lior Pachter
@lpachter
Apropos nothing... I was visiting a friend earlier this year who lives in an apartment in Hyde Park across from where Al Capone used to conduct "business", and the foyer still has a bullet hole in the window from a gunfight that took place in the 1920s.
-
Lior Pachter
@lpachter
This is a good example of why it's important to always set a seed when running a simulation like this. As it is, the result below is not reproducible. Here is reproducible code with a seed.
therealityofthings
@therealityofth1
you’re an idiot
-
Lior Pachter reposted
Michael Baym
@baym
This could just as easily be said about doing science to chase grants
Jeff Weiss
@Passionweiss
Quincy Jones casually offering the greatest quote I've ever heard about the act of creation.
www.vulture.com/2018/02/quincy-jones-in-conversation.html -
Lior Pachter reposted
sina
@sinabooeshaghi
Precision medicine, like precision engineering, requires precise tools. To achieve Formula One-level performance with medicine, we must bring engineering principles to biology. #engineerbiology.
This media couldn't be displayed. See it on the original post. -
Lior Pachter reposted
Shreejoy Tripathy
@neuronJoy
Q: what's more scary than @lpachter calling out your lab's umaps?
A: A "Boo-Map"!
Happy Halloween from TripLab! -
Lior Pachter
@lpachter
There are three plots in this figure and all are terrible at conveying what they are intended to convey.
Dmitry Kobak
@hippopedoid
We updated our ICLR dataset (see github.com/berenslab/iclr-dataset) with blind 2025 submissions to @iclr_conf. Over 10k submissions this year.
I like how this embedding (SBERT + tSNE) shows which ML areas are old-school and which ones are currently booming. -
Lior Pachter reposted
sina
@sinabooeshaghi
The dreaded gene name turns date in Excel strikes again (in a published paper!).. this happens because SEPT1 (Septin 1) is interpreted as September 1st by Excel's automatic date formatting.
-
Lior Pachter reposted
Joel David Hamkins
@JDHamkins
How do you think of the complex numbers? What is the core structure?
1. The complex numbers, at bottom, are an algebraically closed field ⟨ℂ,+,·,0,1⟩.
2. The complex numbers are the algebraic closure of the reals, a distinguished subfield ⟨ℂ,+,·,0,1,ℝ⟩.
3. The complex numbers are an algebraically closed field with a distinguished real coordinate structure ⟨ℂ,+,·,0,1,Re,Im⟩. -
Lior Pachter
@lpachter
I’ve decided not to dress up for Halloween this year.
-
Lior Pachter
@lpachter
Somehow an editor + "peer review" of this 2024 paper in @WileyGlobal's @ImmunityAnd journal approved publication of the trapezoidal rule as "Tai's formula", and a butchered rendering of the formula that is so bad the math is still crying.
Post not available here, view on X
Lior Pachter
@lpachter
Somehow not only were symbols dropped symbols, e.g. 1/2 turned to 12, but Xi and Yi are swapped erroneously. It's very difficult to imagine how the authors (which had written this correctly in a previous paper), an editor, and several "peer reviewers" could produce such a thing.
Lior Pachter
@lpachter
And still people on here trying to argue that the problem with @biorxivpreprint is that papers haven't undergone the "rigorous peer review" of journals ?!?!
-
Lior Pachter
@lpachter
N=1 science is like a waterproof teabag.
Nick Norwitz MD PhD
@nicknorwitz
N = 1 Science is the Future! 💪
-
Lior Pachter reposted
Troy J. Bassett
@3VolumeNovel
This week, I added the 25,000th novel to the database! Been bouncing between the English Catalogue volumes 1 and 4 and both have been slow going.
Love to hear stories from users and their finds!
#victorianstudies #digitalhumanities #bookhistory -
Lior Pachter reposted
Joel David Hamkins
@JDHamkins
I asked a question on MathOverflow about whether there can be a continuous partition of space into circles. mathoverflow.net/q/481197/1946 I strongly suspect not, but so far we can refute only the case in which a linked pair of circles occurs.
-
Lior Pachter reposted
MurawalaLab
@PMurawala
Thank you for your kind words. You may also want to check out our recent BioRxiv that we put out. Here in case of tail regeneration, we find pre-existing resident stem cells. www.biorxiv.org/content/10.1101/2024.01.31.577464v1
-
Lior Pachter reposted
Adam Roberts
@ada_rob
I’m so proud of the updated version of #musicfxdj we developed in collaboration with @jacobcollier, available today at labs.google/musicfx.
Over the past year I’ve spent countless hours experimenting with our real-time music models, and it feels like I’ve learned to play a new type of instrument, in the spirit of what we started @GoogleMagenta to achieve.
It was validating to hear Jacob echo these sentiments the first time he played an early version. With his help, we’ve added new controls, raising the ceiling on what players are able to do with it.
Now anyone can experiment with this new instrument and share what they create with the world.
I'll start things off with my most recent jam session...This media couldn't be displayed. See it on the original post.
Google Labs
@GoogleLabs
It’s here! Thrilled to collab with @jacobcollier on our latest #labsession exploring the magical possibilities of generative music and #musicfxdj. Watch full video: youtu.be/y7gKlzvg8xk
Tune in 10/24 at 5 PM ET as Jacob livestreams a MusicFX DJ sesh on his YT channel.This media couldn't be displayed. See it on the original post. -
Lior Pachter reposted
Gael Varoquaux 🦋
@GaelVaroquaux
Didn't @lpachter write a whole paper to explain not to interpret data reductions like this on genomics?
journals.plos.org/ploscompbiol/article?id=10.1371/journal… -
Lior Pachter reposted
Eli Weinstein
@EliWeinstein6
I'm thrilled to present a new paper with @blei_lab and @lizbwood about estimating the causal effects of T cell receptors in humans. arxiv.org/abs/2410.14127
-
Lior Pachter
@lpachter
Hype or hope? A recently published manifesto published by Thomas Kalil, CIO at @SchmidtFutures and colleagues arxiv.org/abs/2409.11654 envisions "...an AI Virtual Cell as a comprehensive AI framework composed of several interconnected foundation models...". A brief review: 1/🧵
Post not available here, view on X
Lior Pachter
@lpachter
Virtual Instruments: "In the context of UR, Virtual Instruments (VIs) are neural networks that take URs as input and produce a desired output."
Narrator: desired output = $$. ⭐️⭐️⭐️⭐️⭐️ 13/
Lior Pachter
@lpachter
Bayesian: "Under a Bayesian formulation of its
predictive function, the predictions made for cell perturbation outcomes could include an uncertainty score, either implicitly via inference, deep kernels, or
through explicit estimation of the full posterior" ⭐️⭐️ 14/Post not available here, view on X
Lior Pachter
@lpachter
Responsibility: appears only once in a box in a figure. ⭐️ 19/
Lior Pachter
@lpachter
Final comment: would have been nice to have an author contribution section detailing the respective contributions of the 42 authors to this AIVC vision. 20/20
-
Lior Pachter
@lpachter
I’m now in my office @michaelbrundage!
-
Lior Pachter reposted
Heather Machado
@heatherEmachado
PhD and Postdoc positions available! Interested in how our immune system evolves with a growing tumour, and how we can use that to fight cancer? Love data analysis? Applications now being accepted in the Machado Lab (www.machado-lab.org/contact) at the University of Cambridge.
-
Lior Pachter reposted
Trevor Cousins
@trevorcousins_
In this preprint, Regev Schweiger, Richard Durbin and I discuss ancient (>5 million years ago) hominin demography
www.biorxiv.org/content/10.1101/2024.10.17.618932v1 -
Lior Pachter
@lpachter
Barabasi's "network medicine" papers now featured on @PubPeer: liorpachter.wordpress.com/2014/02/10/the-network-nonsense…
-
Lior Pachter reposted
-
Lior Pachter reposted
Yisong Yue
@yisongyue
It's always painful to see a community dedicate time & energy trying to make sense of an overhyped sloppy result. Just a huge waste of time in most cases, where one side acts in good faith while the other spews quite a bit of nonsense (and gets rewarded with praise from those not paying close attention to the details).
-
Lior Pachter
@lpachter
Kind of gross that people are mocking the massive geographic bias in genetics and genomics studies. As of 2019, 71.8% of participants were recruited from only three countries: the US, UK, and Iceland.
www.nature.com/articles/s42003-018-0261-x -
Lior Pachter reposted
Saket Choudhary
@saketkc
Indian Olympiad Qualifier in Mathematics (IOQM) 2024 shows an interesting (and potentially suspicious) trend of over-inflation of consecutive roll numbers qualifying the exam.
-
Lior Pachter reposted
Vahid Shahrezaei
@VShahrezaei
Congratulations to @DVolteras for successfully passing his PhD viva. It has been a pleasure to co-supervise this exciting PhD with Philipp Thomas. Thank you @lpachter and Mauricio Barahona for the examination. Looking forward to Lior's Mathematics Colloquium today @ImperialMaths.
-
Lior Pachter reposted
Eric Kernfeld
@ekernf01
Today, I'm excited to present the second big chunk of my Ph.D. work. We are building a thorough assessment of the burgeoning field of gene expression forecasting. Something is deeply wrong.🧵
- Preprint: www.biorxiv.org/content/10.1101/2023.07.28.551039v2
- Code: github.com/ekernf01/perturbation_benchmarking?tab=readme-… -
Lior Pachter reposted
Nicholas Boffi
@nmboffi
very excited to announce that our math department at CMU has an open tenure-track position at the intersection of math and ML!
CMU math is a fantastic environment for research in this area with world-class ML and CS departments next door.
please apply!
www.mathjobs.org/jobs/list/25418 -
Lior Pachter reposted
Andy Clark
@AndyClarkCornel
New opening for an Assistant Professor position in Human Genetics/Genomics at Cornell! academicjobsonline.org/ajo/jobs/28554
-
Lior Pachter reposted
Saligrama lab_WUSTL
@SaligramaLab
Current thinking (mine, a least)....simply absurd....produces pretty figures though!
-
Lior Pachter
@lpachter
I'd previously heard the number 84%.
Itai Yanai
@ItaiYanai
Doing good science is 90% finding a science buddy to constantly talk to about the project.
-
Lior Pachter reposted
Nicky Mulder
@Nicky_Mulder
I am pleased to announce the establishment of an African Bioinformatics Institute thanks to funding from Wellcome Trust and CZI. We seek stakeholder input to shape the ABI to meet the needs of the bioinformatics and user communities. Join us for engagement sessions to contribute.
-
Lior Pachter
@lpachter
The @NIH has a notice posted on "interest in diversity", and yet...
commonfund.nih.gov/TRA/recipients
Lior Pachter
@lpachter
Link to @NIH's statements on diversity, which includes the quote
"The diverse group almost always outperforms the group of the best by a substantial margin."
extramural-diversity.nih.gov/diversity-matters -
Lior Pachter reposted
Jalees Rehman
@jalees_rehman
UMAPs have an interest in accurately visualizing single cell data, and yet.....
-
Lior Pachter reposted
Danielle Wiggins
@dl_wiggins
Hopping back on here briefly to share:
The Division of the Humanities and Social Sciences at the California Institute of Technology (Caltech) invites applications for one tenure-track position at the assistant professor rank in Asian history... applications.caltech.edu/jobs/history -
Lior Pachter reposted
Karin Prien
@PrienKarin
Stolpersteine aus dem Boden stemmen? Mein Gott, das sind das nur für verirrte, kranke Seelen.
WELT
@welt
Unbekannte reißen alle Stolpersteine in Zeitz heraus www.welt.de/politik/deutschland/article253895196/Sachsen-…
-
Lior Pachter
@lpachter
In my Bi 1c class (introduction to biology through the algorithmic lens) I teach the Hinton-Sejnowski papers in Lecture 7 (where the assigned reading is Hopfield's PNAS paper). docs.google.com/presentation/d/1RJ3YISc3yrzrHB_2UIXg9C63V…
Interesting to see the three of them share a Nobel Prize this morning!
Lior Pachter
@lpachter
BTW, the class is "over-medium"... which means that student are assigned reading ahead of time (in this case Hopfield's paper) which one student presents, reviews, and critiques, there's a short lecture, and then work on a question (slide 10) just to play with the ideas a bit.
-
Lior Pachter reposted
Dvir Aran
@dvir_a
Nobel prize for t-SNE!
The Nobel Prize
@NobelPrize
BREAKING NEWS
The Royal Swedish Academy of Sciences has decided to award the 2024 #nobelprize in Physics to John J. Hopfield and Geoffrey E. Hinton “for foundational discoveries and inventions that enable machine learning with artificial neural networks.” -
Lior Pachter reposted
Yad Vashem
@yadvashem
#worldsmileday
"My beloved and most precious child,
... I am taking with me your beloved image ... the rhythm of your breathing, your smile and your crying."
Read Sara Gerlitz's moving letter to her daughter Dita when they parted during the Holocaust
www.yadvashem.org/yv/en/exhibitions/righteous/florczak.asp -
Lior Pachter
@lpachter
There are three things I noticed regarding this case that have not been mentioned much.
First, the @HHS_ORI issued a case summary on this fraud: ori.hhs.gov/node/3871/printable/print
It is shocking that the fraudster, Gian-Stefano Brigidi can continue to submit grant applications. 1/
Calli McMurray
@callimcflurry
Science is built on trust. What happens when someone destroys it?
My first feature for @_TheTransmitter investigates the emotional and existential fallout for one lab in the aftermath of a misconduct case.
www.thetransmitter.org/science-and-society/a-scientific-f…
Lior Pachter
@lpachter
He attempted to defraud taxpayers out of millions of dollars and successfully stole more than $1.4 million. There should certainly be a path to rehabilitation and redemption in the future, but to just continue to allow him to submit grants, even if under supervision, is crazy. 2/
Lior Pachter
@lpachter
Second, people confuse heartiness with forthrightness. Just because someone is fun at the bar, doesn't mean that they are truthful or transparent at work. It's clear that Brigidi's friendships and rapport with many helped shield him from critique of his work. 3/
Lior Pachter
@lpachter
Third, genomics always gets thrown under the bus. This exchange is depressing to listen to.
"I don't care if someone did an experiment of RNA-seq and didn't find your RNA of interest there..." 👀
www.youtube.com/watch?v=ag5JrBslqgU&t=3133s 4/4 -
Lior Pachter reposted
Timothy Gowers @wtgowers
@wtgowers
Needless to say, with graphs of that size, a big computer search was used.
Except that in fact they did not use a computer search -- the graph was found by hand. There is a very interesting section of the paper about earlier failed attempts to use computers for this. -
Lior Pachter reposted
Blainey Lab
@BlaineyLab
Available now via @biorxivpreprint is our preprint on live-cell transcriptomics with engineered virus-like particles @broadinstitute www.biorxiv.org/content/10.1101/2024.10.01.616098v1
-
Lior Pachter reposted
Bgee database
@Bgeedb
Do you know gget by the Pachter lab? You should! It now includes efficient querying of Bgee in Python. Get high quality curated gene expression data directly in Python or command line. #rnaseq #biocuration #python #scrnaseq pachterlab.github.io/gget/en/bgee.html @lpachter @ISBSIB @marc_rr
-
Lior Pachter
@lpachter
What a great plot. @YunDeng5, @ras_nielsen and @yun_s_song show that the claim that there were only 1,280 humans 900,000 years ago is (unsurprisingly) bollocks. www.biorxiv.org/content/10.1101/2024.10.01.615851v1
-
Lior Pachter reposted
Dr. Heidi Smith Parker
@HeidiKayDeidi
Congrats!!!! @Pselaphinae Joe Parker Named MacArthur Fellow - www.caltech.edu www.caltech.edu/about/news/joe-parker-named-macarthur-fellow.
-
Lior Pachter reposted
sina
@sinabooeshaghi
Thrilled to have assisted in producing this podcast about the standardization of genomics assays with "seqspec" (seek-speck). Tune in 🔊! Video coming soon :)
doi.org/10.1093/bioinformatics/btae168This media couldn't be displayed. See it on the original post. -
Lior Pachter
@lpachter
Bold claims but where’s the proof with the UMAP?
Ben Siranosian
@BenSiranosian
I trained a LLM for single cell RNAseq generation recently. Terabytes of data. Megawatts of GPU usage. Lights in houses dimmed all throughout silicon valley.
In the end, it learned to predict cell type. As in, the predictions were only slightly better than the mean expression! -
Lior Pachter reposted
bioRxiv Bioinfo
@biorxiv_bioinfo
Cell-type-specific mapping of enhancers and target genes from single-cell multimodal data biorxiv.org/cgi/content/short/2024.09.24.614814v1 #biorxiv_bioinfo
-
Lior Pachter reposted
Laura Luebbert, PhD
@NeuroLuebbert
Just in time for my 29th birthday today, I am so excited to present: gget v0.29.0 🎉
We added four new modules that are all about
- gene-disease effects
- cancer
- drugs
- therapeutic targets
🧵 -
Lior Pachter reposted
Jonathan Jacobs 🧬🦠🕹️🏕️
@bioinformer
I’d love to see @bjork talk about cells being like a city… youtube.com/clip/Ugkxvd_13spG5zjjdSkhjNivIjBp-YpUAJ5l
Lior Pachter
@lpachter
A cell is like a city. A cell is a bag of RNA. A cell is a special snowflake. Cells can be organized in a periodic table.
I don’t find these metaphors and analogies to be very useful. A cell is a cell is a cell. -
Lior Pachter reposted
Kentsis Research Group
@KentsisResearch
I've been railing against the use of metaphors in science for a long time! www.nature.com/articles/nature04901
-
Lior Pachter
@lpachter
A cell is like a city. A cell is a bag of RNA. A cell is a special snowflake. Cells can be organized in a periodic table.
I don’t find these metaphors and analogies to be very useful. A cell is a cell is a cell. -
Lior Pachter
@lpachter
This is disgusting and stupid. What a loser.
Nassim Nicholas Taleb
@nntaleb
Fucking ret8rd, we option traders have routinely used negative probabilities to match a payoff.
-
Lior Pachter reposted
Leo C. Stein is @duetosymmetry on bsky/threads/🐘
@duetosymmetry
Is Taleb serious? I went to Caltech with Ben, he's a freaking genius. The required core classes there were probably like getting through junior year in physics, chem, and math majors at most other places.
Nassim Nicholas Taleb
@nntaleb
One thing economists like @ben_golub share, is a deadly combination of confidence and scientific ignorance.
-
Lior Pachter reposted
David Bessis
@davidbessis
"Until 2017, my vision was 99% nonsense. Now it's 96% nonsense. This proves we're on a journey to the Moon."
Seriously, you can't make this stuff up. -
Lior Pachter
@lpachter
I’m curious to know how this language model was trained. It’s nontrivial to make something that generates such gobbledygook.
Joseph Noel Walker (in Washington DC 12-17 Nov)
@JosephNWalker
What is probability really?
Nassim Taleb (@nntaleb) explains to me how he understands it:This media couldn't be displayed. See it on the original post. -
Lior Pachter reposted
Sasha Gusev
@SashaGusevPosts
There was an interesting discussion earlier in the week as to whether a linear mixed model can provide a calibrated test for selection in the presence of drift alone. This is actually pretty straightforward to simulate to let's take a look: 🧵
Sasha Gusev
@SashaGusevPosts
I think there are two interesting questions here: (1) If you just have evenly sampled data from neutral populations, does the GRM capture drift well enough to produce a calibrated test. I.e. if you simulated data from two drifted populations and put it in the model, would you ...
-
Lior Pachter
@lpachter
I agree with this statement from www.biorxiv.org/content/10.1101/2024.09.16.613342v1 "... the application of transformers to single-cell data is currently based on an analogy of genes as words and cells as sentences; yet, such theoretical arguments are not sufficient.."
-
Lior Pachter reposted
Nature Computational Science
@NatComputSci
📢Out now! @GorinGennady, @lpachter and colleagues from @Caltech propose a method to cluster cells from multimodal single-cell data under a self-consistent, biophysical model. www.nature.com/articles/s43588-024-00689-2
-
Lior Pachter reposted
Shriram Krishnamurthi (primary: Bluesky)
@ShriramKMurthi
9/ Basta! Enough! This is like a recipe that first spends three paragraphs on the author's childhood and travels. I know why you're really here: you just want to see the grades! ↵
-
Lior Pachter reposted
Tim van Erven
@tverven
The final draft of @BachFrancis's excellent new learning theory book is available: www.di.ens.fr/~fbach/ He doesn't cover VC dimension, but builds up from regression to general smooth losses, which is heresy of course, but actually fills a much-needed niche. Highly recommended!
-
Lior Pachter
@lpachter
When the UMAP sends you a message.
-
Lior Pachter reposted
NimwegenLab
@NimwegenLab
No point repeating myself: x.com/NimwegenLab/status/1761549964826550438
btw. I wouldn't mind if people only used them for exploration and hypothesis generation in the privacy of their own office. But that is NOT what is going on. They are presented as *results* in virtual every work in the field. 1/1
NimwegenLab
@NimwegenLab
I see we are getting to the stage of the discussion where people are starting to defend UMAP saying it can 'reveal patterns' or 'structure' in the data. Without ever specifying what precisely these patterns/structures represent. This is not surprising because hardly anybody 1/n
-
Lior Pachter
@lpachter
This is insane.
In the UMAP plot below, the blue pair and the red pair are exactly the same distance apart. So in what sense does that "10" make ANY sense for a UMAP plot in @hippopedoid's UMAP?!?
Dmitry Kobak
@hippopedoid
Wow, I did not realize it's in the Nature checklist! :-/
In our plots, to emphasize "the space" aspect, we sometimes add a scale bar (see example), similar to how it's done on actual maps. Not sure if it's needed in actual applications, but I like it. -
Lior Pachter
@lpachter
In @Josephmrich et al. www.biorxiv.org/content/10.1101/2024.04.04.588111v2, we assessed the Seurat and Scanpy single-cell packages, finding that they produce very different results on standard analyses. A new paper in @eLife looking at clustering finds the same elifesciences.org/reviewed-preprints/98469
-
Lior Pachter
@lpachter
This is interesting. A survey by some MIT folks, including VC types, ostensibly surveying student wellness. I wonder if they are harvesting data to throw into some AI with the hope of monetizing the results. Why else the secrecy about the intent?
docs.google.com/forms/d/e/1FAIpQLScJ02VKqpYb50IftCCa0sm4L… -
Lior Pachter
@lpachter
This week I had to make a UMAP for a dataset I'm analyzing. It's interesting... after staring at tons of UMAPs from lots of different tissues... one can kind of tell what one is looking at just from the shape.
Can you guess where this data is from? -
Lior Pachter
@lpachter
This resonated with me.
It's the speaker who looks at me and laughs that they shouldn't show UMAPs because Lior Pachter is in the audience... and then proceed to show dozens of UMAPs in their talk.
Oded Rechavi
@OdedRechavi
Presenting a poster on something that you plan to patent
-
Lior Pachter reposted
Nature Biotechnology
@NatureBiotech
The Wild West of spike-in normalization go.nature.com/3XLUMw3
-
Lior Pachter
@lpachter
Reading a well-written thesis is a delightful experience. Highly recommend ⭐️⭐️⭐️⭐️⭐️
-
Lior Pachter reposted
Anirban Maitra
@Aiims1742
Philanthropist Emmet Stephenson establishes the Stephenson Prize in #pancreaticcancer Research at @cityofhope with a $150 million gift that will fund pancreatic cancer researchers in the US & around the world to transform the landscape of this disease.
www.latimes.com/business/story/2024-09-17/city-of-hope-ge… -
Lior Pachter reposted
Prof. Nikolai Slavov
@slavov_n
Reminds me of "alternative facts".
-
Lior Pachter
@lpachter
This article completely misses the point. A major issue in biology right now is that engaging in data generation does not demand a substantial time investment in bioinformatics research. www.cell.com/cell/fulltext/S0092-8674(24)00839-0
-
Lior Pachter
@lpachter
In 1991 I visited @youngsdairy and was pretty shocked to see that they were selling raw milk. They continued to do so until 2023 when they were finally stopped after 55 cases of Salmonella were connected to the dairy. www.farmanddairy.com/news/raw-milk-sales-in-ohio-end/1559…
-
Lior Pachter
@lpachter
Four years ago an Israeli researcher claimed he'd found a cure for cancer (and the key to fighting Alzheimer's disease).
And yet, here we are.
www.haaretz.com/magazine/2020-11-21/ty-article-magazine/.… -
Lior Pachter reposted
Uri Manor
@manorlaboratory
@lpachter must have made it into OpenAI’s training data one way or another 💀💀💀
Stephen Turner
@strnr
We're cooked.
-
Lior Pachter reposted
Shital Shah
@sytelus
Terence Tao’s grading:
GPT-4o: Completely incompetent graduate student
o1-preview: Mediocre but not completely incompetent graduate student
A step change. -
Lior Pachter
@lpachter
Going to be an interesting evening.
-
Lior Pachter
@lpachter
Not surprised to see the points in this UMAP trying to escape out of the plot they've been forced into.
Source: www.nature.com/articles/s41591-024-03214-0
Lior Pachter
@lpachter
In this UMAP, there are arrows linking nothing to nothing (see panel d). Gornisht mit gornisht, as they say.
Also drawing curves on top of a UMAP built from x-ray images by repurposing RNA velocity software that already didn't make sense is next level!
Lior Pachter
@lpachter
Adding to this thread:
Gus Hamilton
@gushamilton
Wow. Clearly, causal EHR inference has a long way to go (www.nature.com/articles/s41591-024-03214-0). Why on earth would ceph/pen increase LoS but carbapenems decrease them? Does anyone peer review at Nat Med? I mean, this is literally just a dump of data that makes *no* sense
-
Lior Pachter reposted
Gus Hamilton
@gushamilton
Wow. Clearly, causal EHR inference has a long way to go (www.nature.com/articles/s41591-024-03214-0). Why on earth would ceph/pen increase LoS but carbapenems decrease them? Does anyone peer review at Nat Med? I mean, this is literally just a dump of data that makes *no* sense
-
Lior Pachter
@lpachter
I heard a great talk today by @mwpen on work with @AlexCope241 and @jgschraiber that is described in this @biorxivpreprint: www.biorxiv.org/content/10.1101/2024.07.08.602411v2.abstract
Coupled SDEs for modeling the joint evolution of (steady-state) protein and mRNA abundance. Super interesting. -
Lior Pachter
@lpachter
I really don't understand what is going on. How are journals and reviewers ok with this kind of junk? The many organizations that funded this work should be outraged. And sorry for the patients whose data is trashed in this way. Scientists really ought to have some self respect!
Lior Pachter
@lpachter
In this UMAP, there are arrows linking nothing to nothing (see panel d). Gornisht mit gornisht, as they say.
Also drawing curves on top of a UMAP built from x-ray images by repurposing RNA velocity software that already didn't make sense is next level! -
Lior Pachter
@lpachter
So this plagiarism thing has happened to our lab.. again. This time it's plagiarism of our poseidon syringe pump paper @booeshaghi et al., 2019 in @SciReports: www.nature.com/articles/s41598-019-48815-9
Text has been plagiarized, as well as figures copied directly here: www.ijirset.com/upload/2024/march/160_An.pdf 1/🧵 -
Lior Pachter
@lpachter
I respectfully disagree. I think what is currently "democratizing" science much more than "high impact" journals is @PubPeer. So let's take a tour of the @PubPeer comments for some of this author's papers in "high impact" journals: 1/
Iannis Aifantis
@iannisaifantis1
I'll go further Itai..."glam" or actually "high impact" journals are actually very much "democratizing" science. If you take them out everything will be based on pedigree, "fame", X-followers etc...so i would be very skeptical with the idea of getting rid of journals...
-
Lior Pachter
@lpachter
I suspect this is a huge problem, that is being greatly underestimated / ignored.
Elisabeth Bik
@MicrobiomDigest
AI can be used to make convincing looking fake data.
There is no way we can distinguish real from fake anymore.
It is safe to assume that genAI has already resulted in many fake papers. Scientific publishing is not prepared for this.
Cheshire
@Thatsregrettab1
Generative AI fraud example by me. The rightmost column of tumors was created by AI using the first 6 (supposedly) real ones. Time required: 30 seconds.
-
Lior Pachter
@lpachter
Pierre Cartier died yesterday (h/t @monsoon0). His rumination on Gorthendieck, mathematicians, and mathematics, is worth a read: inference-review.com/article/a-country-known-only-by-name
-
Lior Pachter
@lpachter
Is the #icml2024 acronym being changed to International Conference of Men Learning?
Oumar Kaba
@sekoumarkaba
Seems like #icml2024 understimated people's interested in geometric deep learning. So we had to make the @GRaM_org_ workshop overflow in the corridors to manage the great turnaround! 🙌🏾
-
Lior Pachter
@lpachter
I really enjoyed reading this article by Jeremy Avigad and highly recommend it. Thanks @monsoon0.
P.S. It's interesting to ponder similar questions for biology.
Nalini Joshi
@monsoon0
“What makes mathematics worthy of publication in a journal like the Annals of Mathematics?” Jeremy Avigad in fine form www.ams.org/journals/bull/2022-59-01/S0273-0979-2021-0172…
-
Lior Pachter
@lpachter
My annual reminder to please stop using Tophat. Still papers using Tophat v1 (!) from 12 years ago 😱
There have been many better alternatives around for a long time now. Screenshot below is from www.nature.com/articles/s41391-023-00679-x -
Lior Pachter
@lpachter
The edgeR differential analysis tool has been updated to version 4.0., and this update features support for isoform-level DE, which is important functionality that can be used for #scrnaseq (via pseudobulk). Great to see that transcript-level analysis has become mainstream. 1/🧵
bioRxiv Bioinfo
@biorxiv_bioinfo
edgeR 4.0: powerful differential analysis of sequencing data with expanded functionality and improved support for small counts and ... biorxiv.org/cgi/content/short/2024.01.21.576131v1 #biorxiv_bioinfo
-
Lior Pachter
@lpachter
This is disgusting. Shame on the reviewer who wrote this. Shame on the @NIH for sending this kind of summary to the PI.
Mark Moehle Lab
@LabMoehle
Grant unfortunately not discussed. This was at least comic relief when reading the summary statement. I guess connecting flights aren't a thing for the best scientists. Should have pushed for a GNV to SFO flight in my offer letter.
-
Lior Pachter
@lpachter
Is this a math article or an NIH grant proposal?
-
Lior Pachter
@lpachter
This past April I visited Germany to go to a conference in Bonn, and to present several lectures at two Max Planck Institutes. In between I visited several sites of synagogues burned down on Kristallnacht, 85 years ago today. These "Ehemalige Synagoge" are all over Germany. 1/
-
Lior Pachter
@lpachter
-
Lior Pachter
@lpachter
Thinking of my graduate school classmate Danny Lewin today. He was likely the first victim of the September 11, 2001 attacks, just about this time 22 years ago. en.wikipedia.org/wiki/Daniel_Lewin
-
Lior Pachter
@lpachter
Israel: hold my beer.
-
Post not available here, view on XPost not available here, view on X
Lior Pachter
@lpachter
To be clear, I'm not criticizing @GaryMarcus or anyone else for signing this. I'm very curious how this will play out. I just think there is not enough emphasis placed on ideology vs. technology when it comes to technology in general, and GPT in particular.
-
Post not available here, view on X
Lior Pachter
@lpachter
The math of the unitope is rather interesting. In this paper we also discuss how to reverse engineer some information about the raw data (which USNWR does not release) based on the weights and rankings. arxiv.org/abs/0805.1026
-
Lior Pachter
@lpachter
Science has become increasingly exclusionary due to costs of tool subscriptions, conferences & journals. I grew up in South Africa, where faculty starting salaries today are <$2k/month. Participation in the scientific community is increasingly out of reach for most of the world.
-
Lior Pachter
@lpachter
Mind boggling to consider that Tom Lehrer, who is 94 years old, retired from writing and performing music... 50 years ago (!)
He has placed all of his songs in the public domain on this website: tomlehrersongs.com -
Lior Pachter
@lpachter
I had the pleasure & privilege a few days ago of hearing @VikingurMusic perform Ravel's Piano Concerto in G in his debut with @nyphil (Stéphane Denève conducting). A superb performance grounded in a sublime rendition of the 2nd movement and complemented with a beautiful Roussel.
-
Lior Pachter
@lpachter
The first @Bioconductor release of Voyager by @LambdaMoses for spatial transcriptomics is now available. Voyager currently provides support for @10xGenomics' Visium & Xenium, @vizgen_inc's MERFISH, @nanostringtech's CosMX & Slide-seq. Vignettes here: pachterlab.github.io/voyager/index.html 1/
-
Lior Pachter
@lpachter
Despite all the talk about DEI, racists continue to be harbored, tolerated, and promoted in academia. Case in point is this racism from a computer science professor:
-
Lior Pachter
@lpachter
The Trundholm sun chariot (1400 BC) and the Gordieiev bicycle (2022). h/t @SergeBelongie
-
Lior Pachter
@lpachter
(1/9) `ffq` (Fetch FastQ) is a command line tool for finding #sequencing data from public databases like SRA / GEO / @ENASequence / @NCBI / @DDBJ_topics / @ENCODE_NIH. In this quote-tweet thread, @agalvezm, @sinabooeshaghi and I describe the new features.
github.com/pachterlab/ffq -
Lior Pachter
@lpachter
US mass shootings have become so commonplace that sometimes they don't even make front page news. When they do, we learn that "Investigators are still working to determine what led up to the shooting."
You know what led up to the shooting? The distribution of 400 million+ guns. -
Lior Pachter
@lpachter
A bioinformatician speaks truth to power.
KQED Science
@KQEDscience
Mikhail Gelfand’s antiwar letter has drawn hundreds of signatures from Russian scientists.
More via @ScienceMagazine:
www.science.org/content/article/step-nowhere-russian-scie… -
Eric Topol
@EricTopol
About 5 million children age 5-11 now vaccinated with their first dose (nearly ~1 million with 2nd dose) and I've yet to see a case of myocarditis reported 👍
Lior Pachter
@lpachter
Turns out this piece by @wespegden, @sdbaral and @VPrasadMDMPH was wrong. Completely wrong. blogs.bmj.com/bmj/2021/05/07/covid-vaccines-for-children-…
-
Lior Pachter
@lpachter
The Division of Biology and Biological Engineering @Caltech is seeking new faculty in all areas of biological science and biological engineering. Applications will be reviewed starting on 1 November and must be submitted no later than 1 December, 2021. www.nature.com/naturecareers/job/positions-in-biology-and…
-
Lior Pachter
@lpachter
It's time to stop making t-SNE & UMAP plots. In a new preprint w/ Tara Chari we show that while they display some correlation with the underlying high-dimension data, they don't preserve local or global structure & are misleading. They're also arbitrary.🧵www.biorxiv.org/content/10.1101/2021.08.25.457696v1
-
Lior Pachter
@lpachter
The amoral nonsense of Orchid’s embryo selection liorpachter.wordpress.com/2021/04/12/the-amoral-nonsense-…
-
Lior Pachter
@lpachter
My favorite proof that √2 is irrational (Tom Apostol, 2000):
-
Lior Pachter
@lpachter
Fisher may be dead but eugenics ideas are alive and well.
In the video below an MIT prof. explains that he & his students were selected for their "cognitive systems" and that their kids are "genetically selected" for transfer of these "systems".
Harold Pimentel
@hjpimentel
wtf did I just listen to? ah, next level privilege www.youtube.com/watch?v=bgNzUxyS-kQ&t=5488
-
Lior Pachter
@lpachter
Seemingly chaotic and in some cases incoherent recently announced college reopening statements prompted @sinabooeshaghi, @faythtan, @bhrenton, @DrZackaryBerger & myself to analyze #covid19 testing plans for the fall term. www.medrxiv.org/content/10.1101/2020.08.09.20171223v1 1/10
-
Lior Pachter
@lpachter
Black lives matter liorpachter.wordpress.com/2020/06/10/black-lives-matter/
-
Barbara Engelhardt
@BeEngelhardt
JFC. Article on the vision of AI for the future. Eight men mentioned. Seven men quoted. Zero women. Zero people of color. Try again @craigss, and let us know if you need help. www.nytimes.com/2020/04/08/technology/ai-computers-learni…
Post not available here, view on X
Lior Pachter
@lpachter
I can't speak for others but I've tried the personal communication route many times in my career only to be ignored or worse. E.g. even after reporting a sexual harasser confidentially to Title IX, I was encouraged to shut up. So eventually I gave up. liorpachter.wordpress.com/2018/01/17/terry-speed-a-male-f…
-
Post not available here, view on XPost not available here, view on X
Lior Pachter
@lpachter
I'm saying that his "all papers are wrong" paper is wrong and so is his most recent paper which is frustrating because as you point out he is trying to make his own point but then he is failing at that because in fact not all papers are wrong, only his seem to be.
-
Lior Pachter
@lpachter
Diversity matters liorpachter.wordpress.com/2020/01/17/diversity-matters/
-
Lior Pachter
@lpachter
A thread about a landmark reproducibility and usability achievement by @JaseGehring for his @NatureBiotech paper "Highly multiplexed single-cell RNA-seq by DNA oligonucleotide tagging of cellular proteins." The reproducibility repository is here: github.com/pachterlab/GPCTP_2019. 1/19
-
Lior Pachter
@lpachter
So... a colleague at Stanford went and tried to patent *linear regression*. patents.google.com/patent/CA2971129A1/
-
Lior Pachter
@lpachter
-
Lior Pachter
@lpachter
Just read @sean_harrison2's blog post on CCR5-∆32. Interesting, excellent work that is essential reading. seanharrisonblog.com/2019/06/20/ccr5-∆32-is-deleterious-i…
-
Lior Pachter
@lpachter
Happy to announce that we've just posted a preprint on "Modular and efficient pre-processing of single-cell RNA-seq". Highlights: process #scrnaseq on a laptop, 10x processing up to 51 times faster than Cell Ranger. A new efficient RNA velocity workflow. www.biorxiv.org/content/10.1101/673285v1
-
Lior Pachter
@lpachter
Application of an integrative latent variable model that combines factor analysis and an exponential proportional hazards model to TCGA data. Important work by Shannon McCurdy with @AnnetteMolinar2. onlinelibrary.wiley.com/doi/full/10.1002/sim.8151
-
Lior Pachter
@lpachter
"The purpose of models is not to fit the data but to sharpen the questions." —Samuel Karlin
simplystatistics.org/2019/04/17/tukey-design-thinking-and… -
Lior Pachter
@lpachter
I was wrong (part 3) liorpachter.wordpress.com/2018/11/05/i-was-wrong-part-3/
-
Lior Pachter
@lpachter
James Watson in his own words liorpachter.wordpress.com/2018/05/18/james-watson-in-his-…
-
Lior Pachter
@lpachter
The website for Jim Watson's 90th birthday gala is filed under /giving: www.cshl.edu/giving/double-helix-meeting-honor-james-d-wa…
Apparently the event generated enough "giving" (>$750,000) to fund a new endowed chair @CSHL. I fully expect this chair in his honor to be called "The Franklin professorship". -
Lior Pachter
@lpachter
A beautiful essay by Michael Jordan on AI calling for us to "broaden our scope, tone down the hype and recognize the serious challenges ahead." Essential reading medium.com/@mijordan3/5e1d5812e1e7
Lior Pachter
@lpachter
A talk on the same subject www.youtube.com/watch?v=4inIBmY8dQI
-
Lior Pachter
@lpachter
Anat Caspi #womeninscience www.washington.edu/wholeu/2017/12/22/staff-story-anat-caspi/
-
Lior Pachter
@lpachter
Terry Speed: a “male feminist” liorpachter.wordpress.com/2018/01/17/terry-speed-a-male-f…
-
Lior Pachter
@lpachter
Shafi Goldwasser joins @UCB_EECS and is new director of the Simons Institute news.berkeley.edu/2017/10/10/shafi-goldwasser-appointed-d…
-
Lior Pachter
@lpachter
A latent variable model for survival time prediction with censoring and diverse covariates arxiv.org/abs/1706.06995
-
Lior Pachter
@lpachter
Postdoctoral fellowships available in computing and mathematical sciences @Caltech www.mathjobs.org/jobs/jobs/9513
-
Lior Pachter
@lpachter
.@BeEngelhardt speaks on Wed. @UCBerkeley on structured factor models to find interpretable signal in genomic data events.berkeley.edu/index.php/calendar/sn/ccb.html?event_…
-
Lior Pachter
@lpachter
A useful tool for learning some intuition for the behavior of t-SNE distill.pub/2016/misread-tsne/
-
Lior Pachter
@lpachter
"Kuznetsov says, "There is no 'junk DNA'. At least 80 per cent of the human genome initiates RNA production." medicalxpress.com/news/2016-01-junk-dna-center-stage.html
-
Lior Pachter
@lpachter
Straining metagenomics liorpachter.wordpress.com/2015/10/27/straining-metagenomics
-
Lior Pachter
@lpachter
The phylogeny of popular music arxiv.org/abs/1502.05417
-
Lior Pachter
@lpachter
Pythagoras' rule www.dam.brown.edu/people/mumford/blog/2015/Pythagoras.html









































































































































































































































































































