🚀 Our perspective is out in nature! We present a roadmap for Multimodal Foundation Models (MFMs) — large AI models pretrained across multi-omics and multi-timepoint data — to serve as the computational backbone for building virtual cells. Read the full paper in Nature:

Exai Bio presented new data demonstrating the power of its RNA and AI based liquid biopsy platform to detect early stage colorectal cancer. Stage I sensitivity in an independent cohort was 80%, exceeding other blood based approaches. Thanks to Amir Momen-Roknabadi (@aroknabadi)

Check out our latest preprint, led by Siyu Chen, that brings together an amazing team of scientists from several labs to tackle a fundamental problem in biology: how do cells regulate expression of their tRNAs? biorxiv.org/content/10.110…

1/ 🧵 THREAD: Here’s how we solved a 30-year mystery about an HIV drug. 💊 A drug used for over two decades ⚠️ Pulled from the market for cardiac risk 🤯 And no one knew why — until now We found the missing mechanism using our single-cell mosaic dataset, Tahoe-100M. Let’s talk

Fast and Low-Cost Genomic Foundation Models via Outlier Removal 1. GERM is a new genomic foundation model that tackles a key problem in transformer-based genomics models: outliers in attention layers that hinder efficient adaptation and quantization. By removing these outliers,

💥New Publication Alert in Clinical Cancer Research💥 We are thrilled to announce our latest publication on our novel RNA and generative AI liquid biopsy platform for early colorectal cancer detection. In the independent validation set, we achieved an overall sensitivity of 89%

1/8 I’m pleased to share our paper on exai bio’s RNA + generative AI liquid biopsy platform for early colorectal cancer detection: bit.ly/3Fbd0QY #EarlyDetection #LiquidBiopsy

💡 Despite the many clinical AI models being developed, few actually reach real-world deployment due to 𝐡𝐚𝐫𝐦𝐟𝐮𝐥 𝐝𝐚𝐭𝐚 𝐬𝐡𝐢𝐟𝐭𝐬 between training and deployment. Our work, now published in JAMA Network Open, addresses this to enable safer AI for patient care. 👇

So they can identify exact locations for strikes?

checkout our latest paper in Nature Biotechnology. We developed a graph-based algorithm that comprehensively generates non-canonical peptides in linear time. nature.com/articles/s4158…

Got an email from my PhD student today—she had traveled to Tehran just a few weeks ago, right before war was unfolding. No complaints. No requests. Just a quiet, steady project update—sent with the kind of calm strength that left me speechless. I sat there staring at the

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 Abhinav Adduri

Our ENS is vital to gut-brain health. When it breaks down, so does health. Treatments? Almost none. Access to functional human enteric neurons at scale powers research, disease modeling and therapeutic discovery. Grateful to everyone who made this possible.nature.com/articles/s4158…

To my American mentors, colleagues, friends, and neighbors — thank you for making this country feel like home, and happy 4th of July!

Very happy to share that validation of our AMP-Diffusion model that we introduced last year has been published in Cell Biomaterials! 🦠🥳Very thankful for our experimental collaborators César de la Fuente for testing it out! 🙏🐁 📜: cell.com/cell-biomateri… 💻: github.com/programmablebi…

Ever wish you could hit "undo" on disease? 🩺🔄 nature.com/articles/s4155… Most drug discovery asks: what does this perturbation do to cells? But we can also ask the reverse: which perturbations undo a disease signature and move cells back toward health? That's the idea behind

My take on scaling in biology: it is not failing but conditional. Success depends on scaling the right axes like context length, data diversity, tokenization, and architecture. Foundation models matter most where data in biology are sparse.

Pretty sharp take from Amir Momen Roknabadi on the role of AI foundation models in biology. Benchmarks against supervised models and baselines (often done on data from human cells) are useful demonstrations; but the real power of these models is in their zero-shot and few-shot

Fantastic single-cell multi-ome atlas of the developing human lung doi.org/10.1101/2025.0…