Therapeutics Data Commons

Therapeutics Data Commons: Multimodal Foundation for Therapeutic Science, developed @Harvard

[Therapeutics Data Commons @ProjectTDC]

Excited to share our new paper in Nature Chemical Biology @nchembio

AI is poised to transform #therapeutic #science

The Commons is an initiative to access and evaluate #AI capability across therapeutic modalities and stages of discovery 1/4

https://t.co/Mr986dlR1c

[Marinka Zitnik @marinkazitnik]

@GaoShanghua @AdaFang_ @KempnerInst @HarvardDBMI @Harvard AutoScientists automatically produces model cards, experimental logs, failed hypotheses, dead-end registries, and research reports.

The goal is not only to discover better models, but also to document how they were discovered.

[AI Scientists powered by ToolUniverse @ Harvard @ScientistTools]

With the ToolUniverse (TU) CLI, there’s no need to set up an MCP server—agents can instantly access tools with a single command. Paste this to your AI agent to use it now! "Read https://t.co/rFjwcPbalH and update to the latest ToolUniverse. I want to use the tu CLI."

[AI Scientists powered by ToolUniverse @ Harvard @ScientistTools]

If you are at NeurIPS, join us tomorrow for the CUREBench Competition Workshop (https://t.co/GS4pi9KBGo) and see how @ScientistTools is driving large-scale evaluation and global benchmarking of AI models for therapeutics and precision medicine

[Marinka Zitnik @marinkazitnik]

@NeurIPSConf week

Sat, Dec 6
🔹 CUREBench - International Competition on AI agents and reasoning models for therapeutics at scale
https://t.co/HKw4IogWAR @GaoShanghua @ScientistTools @RichardYXZhu @sui67713 @ZKong50693 @xiaorui_su with a keynote by @AziziShekoofeh

Sun, Dec 7
🔹 6th AI for Science NeurIPS Workshop – The Reach and Limits of AI for Scientific Discovery
https://t.co/iGxrLETH2I @AdaFang_
We started this workshop series in 2021, when @AI_for_Science was still niche. It is exciting to see how the community and the field have grown and how much potential there is to transform scientific discovery

Sat, Dec 6
🔹 AI Virtual Cells and Instruments – A New Era in Drug Discovery and Development
https://t.co/NjcVHJ82v4 @_michellemli

We are also presenting many papers throughout the week. Here is the first batch and more to follow throughout the week:

• A scalable data layer of knowledge graph AI: https://t.co/Qah2WmcbaI by Lucas Vittor, @ayushnoori @InakiArango, Joaquin Polonuer

• Multi-agent collaboration in knowledge graph environments: https://t.co/42C3ZzdAXE @ayushnoori @InakiArango
Lucas Vittor, Joaquin Polonuer

• Evolutionary reasoning in protein language models: https://t.co/GC4SM6O76t @YEktefaie

Kudos to all stellar students and many thanks to fantastic collaborators @HarvardDBMI @harvardmed @Harvard @broadinstitute @KempnerInst

[Marinka Zitnik @marinkazitnik]

Highlight of PDGrapher in @thecrimson

https://t.co/VW9BSsvJd7

Great work by @justguadaa Xiang Lin and collaboration with K. Veselkov and @mmbronstein

[Michelle M. Li (李敏蕊) @DrMichelleMLi]

1⃣ more day⌛️ to submit your exciting research to our #NeurIPS workshop on AI Virtual Cells and Instruments!

Submit an extended abstract & join our excellent lineup of presenters! 🌟

Please do not hesitate to reach out if you have any questions!

https://t.co/LR99zSxsuL

[Marinka Zitnik @marinkazitnik]

Update from CURE-Bench at @NeurIPSConf: 524 entrants and 298 submissions! Thank you to the CURE-Bench community!

Working on AI for drug discovery and reasoning in medicine? Your agent belongs here

New teams welcome. Tasks, rules, and leaderboard: https://t.co/HKw4IogWAR #NeurIPS2025 #CUREBench

@GaoShanghua @RichardYXZhu @ZKong50693 @xiaorui_su @CurtGinder

Huge thanks to @cziscience @MilkenInstitute @BiswasFamilyFdn @cziscience Rare as One program @harvardmed @BrighamWomens @HarvardDBMI @KempnerInst @MIT @broadinstitute for making this possible

[AI for Science @AI_for_Science]

Pitch the dataset that could spark the next AI for Science revolution 🚀
The PDB revolutionized structural biology (and even helped win a🏅Nobel Prize in 2024). We’re hunting for the next breakthrough dataset that could unlock similar leaps across science—and we want your idea!

[Marinka Zitnik @marinkazitnik]

📣 Competition launch alert! CUREBench competition at @NeurIPSConf 2025

Start here: https://t.co/HKw4IogWAR

Benchmarking AI for therapeutic reasoning, drug discovery, treatment planning, and therapeutic decision-making

🎯 Track 1: Develop AI models that rely on parametric memory alone
🎯 Track 2: Build AI agents that use external tools and resources
🎯 Evaluation will be done by agentic judges and human disease experts, in collaboration with our partners @cziscience and @harvard
💰 $40,000 in prizes, Starter Kit, Travel Awards, and more
⏰ Entry deadline: October 15, 2025

A big thank you to our partners: @cziscience @MilkenInstitute @BiswasFamilyFdn @cziscience Rare as One program @harvardmed @BrighamWomens

Organized by an amazing team: @GaoShanghua @RichardYXZhu @ZKong50693 @xiaorui_su @CurtGinder Sufian Aldogom, Ishita Das, Taylor Evans, Theo Tsiligkaridis. Big congrats to @GaoShanghua for spearheading this effort

@HarvardDBMI @harvardmed @KempnerInst @MIT @broadinstitute

[Marinka Zitnik @marinkazitnik]

🌍 Excited to open up our global evaluation of AI for drug decision-making and therapeutic reasoning @GaoShanghua

Want to shape the future of therapeutic AI, from understanding existing medicines to developing new treatments for diseases with limited options?

Start here: https://t.co/uCdmJV6waU

How you can participate:

1️⃣ Evaluate TxAgent

Test-drive TxAgent, our AI built for therapeutic reasoning across all drugs since 1939, powered by a universal toolbox of 200+ tools. Assess its reasoning on accuracy, clinical relevance, drug safety, and more, each evaluation takes ~10 minutes.

Learn about TxAgent: https://t.co/tvn5L2vZZ0 and https://t.co/7YXZtmfjBP

2️⃣ Challenge the AI

Submit your toughest therapeutic questions: rare diseases, unmet patient needs, or how to safely extend existing treatments, including combination therapies and personalized treatments. Help build an open library of challenges and shape smarter biomedical AI

@GaoShanghua @ZKong50693 @RichardYXZhu @xiaorui_su @CurtGinder Sufian Aldogom

Thanks to our many partners:
@HarvardDBMI @harvardmed @Harvard @KempnerInst @harvard_data @MIT @broadinstitute @BrighamWomens @MilkenInstitute @BiswasFamilyFdn @cziscience

[Marinka Zitnik @marinkazitnik]

📢 AI-enabled drug discovery reaches clinical milestone

https://t.co/noHOTVFo2a

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 target combination for idiopathic pulmonary fibrosis

What distinguishes this study (in addition to clinical data) is the upstream innovation pipeline

This trial marks a turning point: it affirms a potential for AI to do more than generate molecules faster and cheaper; guide discovery, de-risk development and potentially reshape how we develop medicines

A pertinent question is: why did this AI-generated drug candidate advance to clinical testing when so many others have not?

🎯 Cross-disease target discovery and 'time-machine' setup: AI models trained on past data predicted therapeutic targets years ahead of traditional methods, pinpointing TNIK as a promising target

🔬 Robust biological validation: Integrated multi-omic analyses, network biology, and extensive literature mining rapidly validated TNIK’s biological relevance for fibrosis

⚙️ Chemistry design: Generative AI models designed molecules targeting novel binding sites, prioritized drug-likeness and synthetic feasibility, and proactively optimized pharmacokinetics and potency from early stages

@biogerontology @InSilicoMeds

@HarvardDBMI @Harvard @harvardmed @harvard_data @KempnerInst @broadinstitute

[NeurIPS AI4D3 Workshop @AI4D3]

🔍 Call for Reviewers: AI4D3@NeurIPS 2025

We're seeking experts in AI & drug discovery to review submissions for our NeurIPS workshop.

🗓️ Deadline: May 30, 2025
📍 Workshop: Dec 2025, San Diego
✅ Sign up: https://t.co/VDZIKa6i2I

#AI4D3 #NeurIPS2025 #DrugDiscovery #AI

[Marinka Zitnik @marinkazitnik]

📢 🧬 New preprint!
Can we predict which cancer patients will benefit, before treatment begins? @WanXiang_Shen

Immunotherapy saves lives but many patients don’t respond to treatment, and we still lack reliable tools to predict who will benefit

We introduce COMPASS, foundation AI model for immunotherapy response prediction across cancers and treatments

https://t.co/CniHfrrtCW
https://t.co/Ftty4ZfBYi
https://t.co/vFHtRYUJuh

@HarvardDBMI @harvardmed @KempnerInst @harvard_data @broadinstitute @Harvard

Thanks to incredible team @WanXiang_Shen Thinh H. Nguyen @_michellemli @YepHuang @IntaeMoon Nitya Nair Daniel Marbach

🧵👇

[Kempner Institute at Harvard University @KempnerInst]

New in the Kempner's Deeper Learning blog: @AdaFang_ and @marinkazitnik introduce ATOMICA, a geometric foundation model that enables reasoning about biological systems with atom-level precision.

https://t.co/CD03DhMt6A #AI #ML #MedTech

[Ada Fang @AdaFang_]

📄 Preprint: https://t.co/D7ydckVXRt
🚀 Open source code, weights, and data: https://t.co/gng9m4Uvus

[Ada Fang @AdaFang_]

ATOMICA characterizes interesting clusters of putative bacterial zinc fingers and cytochrome proteins.

We're working on getting some of these validated in the lab 🧫👩‍🔬. Stay tuned!

[Ada Fang @AdaFang_]

Using masked token accuracy to proxy representation quality, we see training of ATOMICA follows scaling laws where representation quality improves with increasing biomolecular data modalities 📈

[Ada Fang @AdaFang_]

ATOMICA builds multi-scale representations at the atom, block (amino acid / nucleotide / common chemical motif), and interaction complex scale.

💡 The key is capturing *interaction complexes* - to learn patterns fundamental to chemistry, such as hydrogen bonds & pi-pi stacking.

[Ada Fang @AdaFang_]

Introducing ATOMICA 💫

A model to universally represent molecular interactions (for proteins, nucleic acids, small molecules, and ions) at an all-atom scale 🧵

[DBMI at Harvard Med @HarvardDBMI]

DBMI's @zakkohane & @marinkazitnik, plus @AdamRodmanMD & other collaborators, on how AI is transforming medicine. “Having an instant second opinion... will change, for the better, the nature of the doctor-patient relationship.” https://t.co/rBRWMoANQu