David van Dijk
@david_van_dijk
Assistant Professor @Yale @YaleMed @YaleCSDept | ML/AI comp bio
What if LLMs could “read” & “write” biology? 🤔 Introducing C2S‑Scale—a @Yale + @GoogleAI @GoogleDeepMind collab: we scaled LLMs (up to 27 B!) to analyze & generate single‑cell insights by turning transcriptomes into text 🧬➡️📝 🔗 Blog: research.google/blog/teaching-… 🔗 Preprint:…
📢New conference where AI is the primary author and reviewer! agents4science.stanford.edu Current venues don't allow AI-written papers, so it's hard to assess the +/- of such works🤔 #Agents4Science solicits papers where AI is the main author w/ human advisors. 💡Initial reviews by…
📢 AI-enabled drug discovery reaches clinical milestone rdcu.be/eugUu Few AI-designed drug candidates have gone beyond in silico benchmarks. Now, a study in @NatureMedicine @biogerontology reports a successful phase 2a trial of rentosertib, an AI-discovered drug and…
How close are we to simulating living cells? Today @arcinstitute is launching the Virtual Cell Challenge (VCC), an annual competition for evaluating progress towards a virtual cell. Read our @CellPressNews commentary
Cells are dynamic, messy and context dependent. Scaling models across diverse states needs flexibility to capture heterogeneity Introducing State, a transformer that predicts perturbation effects by training over sets of cells Team effort led by the unstoppable @abhinadduri
Exactly what we aimed to enable by open-sourcing Tahoe-100M: foundational work on predicting how cells function in different contexts. State, by our friends at @arcinstitute is the first of many to come. Congrats @yusufroohani @abhinadduri, @genophoria , @davey_burke et al.
Introducing Arc Institute’s first virtual cell model: STATE
Honored to receive the @NSF CAREER Award for my project on Neural Operator Learning for Biomedical Discovery! A huge thank you to my students, collaborators, mentors, and @Yale for their unwavering support! #NSFCAREER @YaleCSDept @YaleEngineering @YaleCardiology @YaleMed…
📢 New AI for rare disease diagnosis: SHEPHERD shows how simulation + knowledge-grounded AI = deep learning for ultra‑low label domains nature.com/articles/s4174… This may shorten the journey from phenotype to diagnosis. Excited to help clinicians leverage fewer labels + more…
An AI agent upgraded its own tools and doubled its bug-fix score. Darwin-style search plus Gödel-style self-reference cracked coding tasks. Pass rate jumps from 20 % to 50 % on SWE-bench-Verified Darwin Gödel Machine (DGM) is a coding agent that rewrites its own code, tests…
Amazing! This is a landmark paper in BioAI! I’ve been waiting for a model like BioReason to bridge DNA foundation models with LLMs. Congrats to @BoWang87 and the big team of collaborators! 🎯 What this truly means: This enables AI systems to provide mechanistic biological…
Thank you @YaleEngineering for highlighting our work!
A universal grammar of biology? Yale Engineering, @YaleMed & @GoogleDeepMind developed Cell2Sentence – a tool that lets large language models "read" cellular data. A breakthrough for drug discovery, disease modeling & more. 🔗: loom.ly/FnEkTHk #YaleResearch
>700K people die each year due to S. aureus infection. Today we show that our AI designed new molecule, synthecin, stops drug-resistant S. aureus MRSA in mouse model💊 We created synthecin w/ SyntheMol-RL, our new RL generative AI. All open source biorxiv.org/content/10.110…
Excited to see our Cell2Sentence collaboration with @GoogleAI @GoogleDeepMind featured in Nature News! Check it out here: nature.com/articles/d4158… 🧬
🚨Preprint Alert!🚨 Cell classification is one of the most difficult tasks in analyzing spatial data. We present CellTune - a powerful toolkit for accelerating biological discovery in spatial proteomics. 📄 biorxiv.org/content/10.110… 👇🧵1/7
HealthBench from @OpenAI is a critical evaluation benchmark to push AI into real-world healthcare & pave the way for AI doctors. This in turn could have a tremendous impact on human health. We’re now at the beginning of revolution that will transform the future of medicine & save…
Evaluations are essential to understanding how models perform in health settings. HealthBench is a new evaluation benchmark, developed with input from 250+ physicians from around the world, now available in our GitHub repository. openai.com/index/healthbe…
I support this! For the past 35 years, I’ve focused exclusively on human-based science, which has been a very challenging path since the conventional approach of biomedical research almost required using mouse models to secure high-profile publications & grants. Somehow, I…
Today, NIH announced a new initiative to expand innovative, human-based science while reducing the use of animals in research. This aligns with @US_FDA’s initiative to reduce testing in animals. bit.ly/4jYd4T2
Great opportunity for building virtual cells!
Today, NIH announced a new initiative to expand innovative, human-based science while reducing the use of animals in research. This aligns with @US_FDA’s initiative to reduce testing in animals. bit.ly/4jYd4T2
You’ve heard of vibe coding—meet vibe protein design. Click, launch, and watch the binders roll in. That’s what we're unveiling with @AriaxBio today.
🪰By infusing a virtual fruit fly with #AI, Janelia & @GoogleDeepMind scientists created a computerized insect that can walk & fly just like the real thing➡️ hhmi.news/3Rwop0w 🤖Read more about this work, first published in a #preprint in 2024➡️ hhmi.news/4cGAVUW