Valeh Valiollah Pour Amiri

@saramostafavi.bsky.social (@Genentech) & I (@Stanford) r excited to announce co-advised postdoc positions for candidates with deep expertise in ML for bio (especially sequence to function models, causal perturbational models & single cell models). See details below. Pls RT 1/

Disease diagnostics using machine learning of B cell and T cell receptor sequences Clinical diagnosis typically incorporates physical examination, patient history, various laboratory tests, and imaging studies but makes limited use of the human immune system’s own record of antigen ...

Disease diagnostics using machine learning of B cell and T cell receptor sequences

www.science.org/doi/10.1126/...

TL;DR: BCRs ARE ALL YOU NEED!

(Well actually .... keep reading) 1/

This was a really fun collaboration with @amarderstein.bsky.social where we explored some of the interesting relationships between context-specific non-coding variant effects, disease, and evolution using deep learning models of chromatin accessibility in the brain and heart.

Truly upsetting. That this is considered “usual”, as most comments indicate, is all the more disappointing. What will it take to reverse the trend? When will it be considered UNusual when women are not credited for their work?

Excited to share our latest work, published this morning in Nature Biotechnology! 🎉 We developed a hybrid quantum computing-enhanced generative model to design KRAS inhibitors. Check it out here: rdcu.be/d68Fs.

Some fascinating insights along the way:

@anusri.bsky.social first author & developer of ChromBPNet is looking for opportunities in industry in ML for bio/genomics. She is an excellent rigorous scientist (as u can see from the paper). Very strongly recommend her. Plz reach out to her if u have openings. Plz forward.

A month ago we @vevotherapeutics.bsky.social announced that we have generated the largest single-cell perturbation atlas in history, Tahoe-100M. Today, we announce that we will fully open-source Tahoe-100M in Feb, as part of a collaboration with NVidia health to train cell state models.

Very excited to announce that the single cell/nuc. RNA/ATAC/multi-ome resource from ENCODE4 is now officially public. This includes raw data, processed data, annotations and pseudobulk products. Covers many human & mouse tissues. 1/

www.encodeproject.org/single-cell/...

A generative framework for enhanced cell-type specificity in rationally designed mRNAs mRNA delivery offers new opportunities for disease treatment by directing cells to produce therapeutic proteins. However, designing highly stable mRNAs with programmable cell type-specificity remains ...

The first preprint of 2025! Together with Matvei, @halfacrocodile.bsky.social, & our amazing team, we are excited to share PARADE: an AI framework for designing mRNA UTRs with enhanced cell-type specificity & stability. www.biorxiv.org/content/10.1...

Good papers lead to more questions than they answer. That's because (as Blaise Pascal wrote) knowledge is like a sphere: the larger it gets the larger also its surface, which is its interface with the unknown.

This is a monumental amount of work and absolutely worth a read if you do anything in genomics. Not only does the underlying modeling idea make sense but the benchmarking is truly next level.

Very happy to see the preprint is out and that the hype of calling it a foundation model was avoided.

Our ChromBPNet preprint out!

www.biorxiv.org/content/10.1...

Huge congrats to Anusri! This was quite a slog (for both of us) but we r very proud of this one! It is a long read but worth it IMHO. Methods r in the supp. materials. Bluetorial coming soon below 1/

Multiplexed mosaic tumor models reveal natural phenotypic variations in drug response within and between populations Many agents that show promise in preclinical cancer models lack efficacy in patients due to patient heterogeneity that is not captured in traditional assays. To address this problem, we have developed...

Our latest from the indefatigable @thejohnnyyu.bsky.social in collaboration with Weissman and Shokat labs. Meet GENEVA, which enables simultaneous phenotyping and profiling of cancer cell drug responses at scale; both in vitro and in vivo across a variety of models: www.biorxiv.org/content/10.1...

Have a genomics ML model and want to use it to design regulatory elements? tangermeme v0.4.3 is out with a focus on design. Check it out!

- faster (3-10x usually)
- extended documentation
- evolutionary design
- new construct-centric design paradigm
- QOL improvements

github.com/jmschrei/tan...

Check out this systematic benchmark of genome-wide, annotation agnostic DNALMs & strong baseline ab-initio models for biologically meaningful tasks in regulatory genomics 1/

DART-Eval: A Comprehensive DNA Language Model Evaluation Benchmark on Regulatory DNA Recent advances in self-supervised models for natural language, vision, and protein sequences have inspired the development of large genomic DNA language models (DNALMs). These models aim to learn gen...

(1/10) Excited to announce our latest work! @arpita-s.bsky.social, @amanpatel100.bsky.social , and I will be presenting DART-Eval, a rigorous suite of evals for DNA Language Models on transcriptional regulatory DNA at #NeurIPS2024. Check it out! arxiv.org/abs/2412.05430

Innovators Unveiled: Announcing the Databricks Generative AI Startup Challenge Winners! Vevo Therapeutics has been selected as the Grand Prize Winner of the Databricks Generative AI Startup Challenge has selected.

To build disease-relevant models of the cell, we need perturbative data at scale + AI that can efficiently learn from that data. @vevotherapeutics.bsky.social may be the only team that is pushing the boundaries of both. This recognition from DataBricke is a testament to that dbricks.co/3DbIULQ