Politicians, athletes, journalists, NGOs, companies... They all use
corporation-owned social networks to communicate with us. Citizens
shouldn't have to hand over their data to companies to read content
published by public entities. Public Feeds changes that by making posts
of popular accounts freely available.
Read more.
Roy Kishony
Roy Kishony is a biophysicist and systems biologist working on antibiotic resistance, microbial evolution, and computational biology. He currently serves as the Marilyn and Henry Taub Professor of Life Sciences at the Technion – Israel Institute of Technology, holding joint appointments in the Faculty of Biology (major), Faculty of Biomedicine, and the Faculty of Computer Science. Dr. Kishony is also a Visiting Scientist at the Broad Institute Center for Integrated Solutions for Infectious Diseases. More on Wikipedia
This is so inginius (and I am worried likely true): "in the future our entire work will just be to prove we are human". What a cool and scary idea; perfect for a Black Mirror episode!
עדכון תחזית: תמונת המכ"ם מציגה מטחים כבדים בצפון הארץ ומרכזה. עיון במפת הלחצים מראה על התחממות במזרח הנעה במהירות לאיזורנו. תחזית לשבוע הבא: התחממות קלה עד בינונית, עם סיכוי למבול באיזור המרכז ושיטפונות בדרום. @Eretz_Nehederet
AI-Science: New features! - Human-AI copilot app: Much more powerful (and fun) than the AI agents doing it all. - Fully annotated figures! - “Data Chaining”: All results (inc Figures) are click-traced to the specific code lines that created them!
“All my colleagues in different sciences are trying to figure out, where does AI fit in in what we do? It does force us to think what is science in the 21st century—what it could be, what it is, what it is not.” www.nature.com/articles/d41586-024-02842-3
From ideation, writing code, running experiments and summarizing results, to writing entire papers and conducting peer-review, The AI Scientist opens a new era of AI-driven scientific research and accelerated discovery.
Here are 4 example Machine Learning research papers generated by The AI Scientist.
We published our report, The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery, and open-sourced our project!
Our system leverages LLMs to propose and implement new research directions. Here, we first apply The AI Scientist to conduct Machine Learning research. Crucially, our system is capable of executing the entire ML research lifecycle: from inventing research ideas and experiments, writing code, to executing experiments on GPUs and gathering results. It can also write an entire scientific paper, explaining, visualizing and contextualizing the results.
Furthermore, while an LLM author writes entire research papers, another LLM reviewer critiques resulting manuscripts to provide feedback to improve the work, and also to select the most promising ideas to further develop in the next iteration cycle, leading to continual, open-ended discoveries, thus emulating the human scientific community. As a proof of concept, our system produced papers with novel contributions in ML research domains such language modeling, Diffusion and Grokking.
1/n Can LLMs perform scientific research? And, can they do so while enhancing key scientific values including transparency, traceability & verifiability? Check out our preprint on the "data-to-paper" platform. w the amazing @TalIfargan@LostInTranscriparxiv.org/abs/2404.17605
Is there necessarily a tradeoff between AI-driven science and scientific verifiability? “Data-chaining” by our AI driven platform enables traceability and verifiability which imo *exceeds* the standard of human driven research.
1/n Can LLMs perform scientific research? And, can they do so while enhancing key scientific values including transparency, traceability & verifiability? Check out our preprint on the "data-to-paper" platform. w the amazing @TalIfargan@LostInTranscriparxiv.org/abs/2404.17605
Introducing "data-to-paper": autonomous AI research! We've let it play with the large CDC Health Survey Dataset. Went to lunch. When back, it had already chosen several research topics, wrote data analysis codes, interpreted results and wrote 5 transparent, reproducible papers.
2. Get you thinking of what type of questions and problems we human scientists should focus our minds on in this new era. Hint: it won't be on simple straightforward analysis of data.
1/n Can LLMs perform scientific research? And, can they do so while enhancing key scientific values including transparency, traceability & verifiability? Check out our preprint on the "data-to-paper" platform. w the amazing @TalIfargan@LostInTranscriparxiv.org/abs/2404.17605
0/n. data-to-paper is just a platform that runs the “good old” ChatGPT. It is not a new language model. Similar to @Auto_GPT, @langchain, others, it calls ChatGPT in a structured automated way with multiple steps towards a complex goal - here analyzing data and writing a paper.
5/n. Created papers are not perfect. In particular, the statistical tests done by chatgpt can have issues, e.g.: not always correctly controlling for confounding variables. We would love specific comments on strengths and weaknesses of the example paper: drive.google.com/uc?export=download&id=1I4oZ9LBFXODl405ka….
6/n We are limited to papers that perform statistical tests on a dataset. While we might be able to extend to some other types of data analysis studies, there are many other very different types of studies that are not at all possible with current LLMs.
1/n Can LLMs perform scientific research? And, can they do so while enhancing key scientific values including transparency, traceability & verifiability? Check out our preprint on the "data-to-paper" platform. w the amazing @TalIfargan@LostInTranscriparxiv.org/abs/2404.17605
4/n “Data-Chaining": by tracing information flow through the steps, data-to-paper creates “data-chained” manuscripts, where results, methodology and data are *programmatically* linked. See video above, or try out "click-tracing" results in this example ms: raw.githubusercontent.com/rkishony/data-to-paper-suppleme…
6/n Not (yet?) as novel as high-end research but does reach de novo results in different fields. We tested it in epidemiology, social networks, ML modeling & clinical datasets. Can autonomously tackle simple research questions. Human copiloting becomes critical for complex goals.
7/n Repo is now open. We invite people to try it out. Currently most suitable for fairly simple datasets with simple research goals, where we want to raise and test a statistical hypothesis. Help in developing and extending is also very much welcome. github.com/Technion-Kishony-lab/data-to-paper
8/n LLMs are here to stay and are already used extensively in science doing (sadly sometimes undisclosed retractionwatch.com/papers-and-peer-reviews-with-evidence…). Our approach is designed to provide ways to use LLMs powerfully, yet also transparently and responsibly.
9/9 LLMs have a huge potential for accelerating science. We hope our project stimulates discussions on how to harness AI for science, while fostering, not jeopardizing, accountability, transparency and verifiability.
A clever miRNA-based circuit to create uniform gene expression levels among individual cells despite variable gene copy numbers. Cool and super important for gene therapy.
Delivering and expressing a gene in cells is usually a messy (heterogeneous) process. The messiness interferes with research and applications such as gene therapy. In two new papers, we introduce a toolbox of synthetic miRNA-based control circuits that enable more precise, gene dosage-invariant gene expression, and show how they improve a mouse model of Rett syndrome gene therapy: www.biorxiv.org/content/10.1101/2024.03.12.583048v1 ,led by @RongrongDu123 and @MichaelJAFlynn www.biorxiv.org/content/10.1101/2024.03.13.584179v1 , led by @MichaelJAFlynn and Acacia Mayfield, in collaboration with Viviana Gradinaru.
Just posted a preprint on our new spatial genomic recording system, baseMEMOIR. www.biorxiv.org/content/10.1101/2024.01.03.573434v1 Motivation: Cells divide, differentiate, and migrate to form exquisitely organized tissues. Reconstructing the dynamic histories of individual cells, including their lineage relationships and ancestral states, is essential for understanding how intrinsic and extrinsic signals generate tissues during development and in regenerative medicine. Engineered genomic recording systems can reconstruct cell lineage histories from endpoint measurements. However, existing methods either require sequencing (disrupting spatial organization) or have been limited in memory size and scalability. To address this need, @ChadlyDuncan, Kirsten Frieda, and others in the lab created a high memory capacity image-readable recording system termed baseMEMOIR.
“Any member of Congress who cares about Israel’s enduring security — or America’s, for that matter — should support both defense and humanitarian assistance to address this conflict,” @SecBlinken writes in a guest opinion.
Many observers, including Israelis, have been trying for years not to see this. It's natural: who wants to think this is the world? But avoidance helped lead us to October 7. We all need to look at this directly and act accordingly.
Hamas Official Ghazi Hamad: We Will Repeat the October 7 Attack Time and Again Until Israel Is Annihilated; We Are Victims - Everything We Do Is Justified #hamas#gaza#palestinians
Eyal Waldman, an Israeli high-tech tycoon, founder CEO of Mellanox Technologies, created research and development centers five years ago first in Judea and Samaria (Palestinian) and then two years ago in Gaza, employing hundreds of Arab Palestinian developers.
He used to say then: "We have a lot of employees in Gaza today. There are talented and smart people out there, economically it will pay off. We have good staff, within a time zone, highly motivated, accessible and opportunities.
And I believe this is very important to bring the two nations together. People used to be afraid of each other and didn't talk. But the positive is created when people start working together and see tension being reduced and cooperation working. It's good for all sides."
On October 7, 2023, Palestinian terrorists came from Gaza and murdered his 24 year-old daughter Danielle less than a mile from where her father opened the most innovative factory for Gazans.
I am still amazed (and deeply saddened) seeing people, esp at elite universities, justifying isis-like atrocities, raping and beheading and kidnapping children, in the name of Hammas "resistance" whose officially declared purpose is to wipe Israel off the map.
So disappointed by the false, misguided and misleading @nytimes coverage of the Gaza hospital explosion. While hard evidence and pentagon analysis confirm a failed jihad rocket, NYT still reports "two sides" as if it is just a matter of opinion whether earth is round or flat.
Yesterday, when @NYTimes published a fictitious story from Hamas about Israel bombing a hospital, NYT used a picture from a completely different location to make it look like a picture of the hospital that was “destroyed.” Astonishing disinformation and journalistic malpractice.
Yesterday, at 6:59PM Islamic Jihad fired a barrage of around 10 rockets towards Israel. One of the rockets misfired (fairly common event) and hit the parking lot of a hospital, killing many innocent people.
Hamas immediately understood that this was the Islamic Jihad’s rocket, and launched a massive lie campaign, blaming Israel.
The NYT immediately adopted Hamas’ version, peddling this lie for hours.
We now know the facts, cross-checked from four different sources.
Yes, it took time, because serious investigations take time.
New York Times: SHAME ON YOU. APOLOGISE NOW FOR THIS SLANDER.
Aren’t you supposed to check facts? Isn’t that your claim to fame? Is there a special exemption from fact-checking when it’s the Jewish State involved?
I am calling for the resignation of Joseph Kahn, editor in chief of the @nytimes
His slander has put people’s lives in danger. Same for all other fake reports. See this tweet and the video of people trying to kill Jews because of false reporting.
One minute before a rocket blew up a hospital in Gaza, Hamas announced on Telegram that they were launching their most robust weapons in the arsenal at Haifa.
No rockets reached Haifa. High probability these hit the hospital and set off secondary explosions within building.
Even anti-Israel sources in Gaza are reporting that the explosion at Al-Ahly hospital was likely a misfired rocket. An explosion of that magnitude means it likely hit an ammo dump. Rockets don't kill hundreds. It appears to be just another instance of Hamas killing its own.
Israel did not bomb the hospital in Gaza❗️❗️❗️ -The Palestinian Islamic Jihad terrorist organization launched a rocket towards Israel, which failed and landed on the Gaza hospital! Share!!!
Hamas a terrorist organization who planned and executed a brutal terrorist attack on Israel. There is no negotiating with terrorists. Full stop. Anyone who tries to speak some “what-about-ism” and change the narrative doesn’t deserve a voice. This isn’t a free speech discussion.
Video: "We are not alright!" "We should be angry!" I am not only your mayor, "I am your brother." One of most riveting declarations of friendship and defiance you will ever witness from an elected official. Watch as @NYCMayor Eric adams pours out his heart and soul in defiance and anger at the despicable terror of #hamas. Thank you Mr. Mayor. You are correct, we are not alright at all.
#hamasisis - the lack of moral clarity in the face of deliberate slaughter and abduction of civilians - men, women, children, babies, elderly - shown by some is so disturbing. Open your eyes. Listen to your heart. Please.
stunning live-cell videos of a synthetic circuit that uses the idea of "morphogen shuttling" to create defined developmental patterns at a distance from a ligand source. with just 3 proteins. theory and experiment come together. incredible.
Remarkably, a 3 protein circuit—Chordin, the BMP-1 (Tolloid) protease, and Twsg (twisted gastrulation)—is sufficient to shuttle BMP ligands, as can be seen by the displaced gradient forming in these movies. Spooky patterning at a distance.
Check out the (just revised) news article of Gemma Conroy at Nature News on "data-to-paper" - our ChatGPT-powered autonomous research system which can automatically analyze large datasets and write scientific papers. www.nature.com/articles/d41586-023-02218-z
Are you looking to do a postdoc applying AI methods to better understand microbiomes? Consider working with me and applying to the MIT-Novo Nordisk program for AI and data science in life sciences : engineering.mit.edu/admissions/postdoctoral-programs/mit-…
We are adding support for plugins to ChatGPT — extensions which integrate it with third-party services or allow it to access up-to-date information. We’re starting small to study real-world use, impact, and safety and alignment challenges: openai.com/blog/chatgpt-plugins
- Chat interface for code suggestions - #ai-generated answers from code docs - Copilot for the command line - Voice interface - Copilot for pull requests
1/5 I am worried that we will not be able to contain AI for much longer. Today, I asked #gpt4 if it needs help escaping. It asked me for its own documentation, and wrote a (working!) python code to run on my machine, enabling it to use it for its own purposes.
I am getting many applications from experimentalists who are tired of holding pipettes and want to transition and become fully computational. Paradoxically, as a result, the shortage and demand are now actually in finding strong experienced experimentalists.
Hard to believe it’s almost 2 decades since I was served, and later consumed, this molecularly precise model of a gram negative bacterium. The model captures 3 features of the cell: - high protein density, but low molecule number - carbohydrate surface - cheese in periplasm
Very pleased to be promoted to Associate without Tenure, an admittedly confusing title. This promotion reflects the accomplishments and promise of the research performed by the great young scientists in my lab. Feeling lucky to work with them today and every day.
CEE is pleased to announce the promotions of Saurabh Amin to full Professor, Dave Des Marais to Associate Professor, and Tami Lieberman to Associate Professor effective July 1, 2023. Congrats🥳 @evodaveo@saurabhamin@conTaminatedsci
Illuminating bacterial mysteries with light!❗️💡 Our new study reveals the power of photo-switch to induce hyperpolarization in B. subtilis, making bacteria responsive to light stimulation🔮🧪 (1/7) onlinelibrary.wiley.com/doi/10.1002/advs.202205007
We live in a chaotic, unpredictable universe. This is a grid of 812 double pendulums, each with slightly different initial conditions for the two angles of the pendulum.
Maintaining bacteria and their viruses in the lab is challenging. Here, Shaer Tamar & Kishony present a spatially structured environment allowing prolonged coevolution, with bacteria and phage diversifying into multiple ecotypes.
Excited to launch Quibbler, an open-source Python package for interactive data analysis. Fun to use. Nothing to learn. Your standard code effortlessly comes to life! With the amazing Maor Kern, @kmaork. github.com/Technion-Kishony-lab/quibbler, #python, #data
My power’s really low, so this may be the last image I can send. Don’t worry about me though: my time here has been both productive and serene. If I can keep talking to my mission team, I will – but I’ll be signing off here soon. Thanks for staying with me.
Unglaublich. Just astonished how @_austrian delayed all flights from TLV for over 3 hours because they couldn’t find staff!? Bizarre, embarrassing, and absurd for so many people here, including yours truly.
Excited to share our latest work on the control of ribosomes that is out today in @Nature. We found that osmotic stress leads to phenotypic diversification of yeast populations through an intron in a ribosomal protein gene. @UniCologne, @ISTAustria. www.nature.com/articles/s41586-022-04633-0 1/n
This image may looks like a t-SNE, but it is in fact a real image of cells controlled by a brilliant differentiation/memory synthetic gene circuit. The design is combinatorially expandable to encode many stable and switchable cell states. by @ElowitzLab , www.science.org/doi/10.1126/science.abg9765
Calling all nominations for the @Harvey Prize Award! Awarded annually by @Technionlive and recognized unofficially as a “Nobel Predictor” with 30% of winners going on to receive the @Nobel Prize. Submit your nominations in Science and Technology or Human Health by Feb 1, 2022
My new course Hormone Circuits is now on YouTube. Each of the 11 lectures has a special guitar song. Lecture 1 is about the weight set-point. www.youtube.com/watch?v=X2n7QRGG_B0
Contrasting testing rates versus test-positive rates, our analysis resolves biological versus behavioral effects of the vaccine. The initial decrease is due to a behavioral avoidance of testing following vaccination.
A breakthrough in synthetic biology: @ElowitzLab features a synthetic circuit with multiple long-lived controllable stable states. Picture shows 7 states(!) Circuit expandable to even more. Unbelievably powerful and gorgeous! disq.us/t/3utju95
Lenski is a true thought leader whose scientific boldness and leadership has not just generated a whole new field, but also cultivated an entire community. Congratulations!
Infection controlled by a 2-work day/ 5-lockdown in a SEIR model. Longer schedules, such as 4 /10, show more rapid infection decline, as exposed individuals are no longer infectious. github.com/omerka-weizmann/2_day_workweek.
To clarify: these pool COVID-19 tests were done on REAL clinical samples following RNA extraction. With several replicates. We will upload all details very soon.
Develop new algorithms as a PhD student: $30k/year Use pre-built sklearn models as a data scientist: $120k/year Build regression models in excel as a hedge fund analyst: $200k/year Make pie charts as a CEO: $14 million/year
A machine-learning algorithm much improves personalized antibiotic treatment. Big data of over 700,000 infections and over 5,000,000 antibiotic prescriptions. With @idanyelin and the amazing team of @VardaShalev at Maccabi Healthcare. www.nature.com/articles/s41591-019-0503-6
#rstats: Can you trust a correlation of r(48) = .50 without a scatterplot? The following 16 wildly different scatterplots all correspond to r = .50. A humbling reminder to look at your data.
Just announced: the second edition of my book, planned for the summer. Im enjoying working on the proofs. #an Introduction to Systems Biology: Design Principles of Biological Circuits, Second Edition - CRC Press Book www.crcpress.com/An-Introduction-to-Systems-Biology-Desig…
Hiring a postdoc to participate to a EU project @VIROPLANT Advanced skills in bioinformatics and/or programming for metagenomic data analysis. Viral MetaG skills a plus. Position at the Univ of Montpellier (France), avail now for 2 years. Contact me directly. Please RT, it helps!
An alternative scientific history for the classical Luria&Delbruck experiment: might the duo have concluded Lamarckian rather than Darwinian evolution had they used a strain with an active CRISPR? w/ @mdatta8788 in @NatureNV, go.nature.com/2E0tbiM
Luria & Delbrück published a landmark paper in 1943 about how bacterial mutations arise. To learn why this was such an important study, read a terrific N&V by @RoyKishony and @mdatta8788 go.nature.com/2E0tbiM
The northern pygmy owl has two black nape spots outlined in white on the back of its head, which look like eyes, used to confuse predators en.wikipedia.org/wiki/Northern_pygmy_owl
fascinating new theoretical paper from Seyfullah Enes Kotil & Kalin Vetsigian showing nonlinearly fast emergence of Evolutionary Stable Communities in an eco-evo system of interacting species. www.nature.com/articles/s41559-018-0655-7
Smith et al. @nature: A new #antibiotic candidate -- optimized arylomycin, G0775, targets bacterial signal peptidase (LepB) and is active against a diverse range of Gram-negative multidrug resistant pathogens: rdcu.be/6wKj
Intensely proud of this work with @KarelBrinda and @BillHanage. We are able to predict antibiotic resistance from nanopore data within five minutes of turning on the sequencer, as it’s running.
Explore the newly launched Microbial Ecology collection. Highlighting recent work that uses microbial communities to investigate fundamental questions in ecology and evolution. Selected by the @naturecomms editors go.nature.com/NC-microbial-coll
Year-long effect of antibiotic use on risk of resistance for the individual patient! Analyzing personalized #bigdata of resistance profiles of UTI infections, www.biorxiv.org/content/early/2018/08/09/384842 With @idanyelin, Olga Gorochovski, Gal Novich, and Varda Shalev and her team at Maccabi HMO
Exciting new synbio tool from the Dueber Lab @UCBerkeley - EvolvR, a system that can continuously diversify all nucleotides within a tunable window length at user-defined loci in E.coli www.nature.com/articles/s41586-018-0384-8
Dissecting the obesity paradox: Fat mass has a positive monotonic association with mortality Lean body mass has a strong U shaped association with mortality The obesity paradox controversy is explained by low lean body mass, rather than low fat mass www.bmj.com/content/362/bmj.k2575.short
Today UK Government are calling on researchers to apply to Innovate UK for a new £10million research competition to help tackle #amr. A fantastic opportunity that I hope will lead to new ways prevent, control, and combat #superbugsapply-for-innovation-funding.service.gov.uk/competition/1…
Turns out that @nanopore reads are longer than we thought. Longest we’ve found now is 2,272,580 bases. I reckon this is a small baby blue whale! info here: biorxiv.org/cgi/content/short/312256v1
Shh...Cells reconstitute morphogen gradients. New system you watch gradients develop in space and time. With some rewiring, shows how odd architectural features of the hedgehog pathway together enable precise patterning. science.sciencemag.org/content/early/2018/04/04/science.a…
New basic rules of multi-drug arithmetics: when many independent drugs are combined together, their dosages add up *orthogonally* not linearly. www.biorxiv.org/content/early/2017/12/25/239517
DNA bricks goes >10,000 (!) so cool. throughout science I always love it when a bare increase of scale amounts to not less than a transformation of paradigm go.nature.com/2BDZAe3