Come do a PhD internship with me!

You have until Dec 1 to apply to the bioml PhD research internship! This is where you apply to work with me, @alexijie.bsky.social @avapamini.bsky.social @lcrawford.bsky.social or Kristen Severson! (new) link and some instructions below apply.careers.microsoft.com/careers/job/...

Come do a research internship with us! Details on the posting and how to upload a research statement given below!

We have recently uploaded our revised manuscript “Evaluating the role of pre-training dataset size and diversity on single-cell foundation model performance”. TL;DR: More models, more tasks => same results.

Introducing REFLEX, a novel immune profiling assay. This new method allows for the capture of sequence information for potentially millions of T cells simultaneously. 🔗 https://www.biorxiv.org/content/10.1101/2025.10.24.684243v1

We have recently uploaded our revised manuscript “Evaluating the role of pre-training dataset size and diversity on single-cell foundation model performance”. TL;DR: More models, more tasks => same results.

genomebiology.biomedcentral.com/articles/10.... Quite an indictment of some of the current single cell "virtual cell" foundation models. Even for the relatively mundane applications, cell labeling, batch correction etc, they are poor compared to much simpler & cheaper methods.

Great to see our paper presenting recall, a framework which calibrates clustering for the impact of data "double-dipping" in single-cell studies, out today in AJHG! Congratulations, @alandenadel.bsky.social and co-authors!

🚨Online now! 📄Artificial variables help to avoid over-clustering in single-cell RNA sequencing 🧑‍🤝‍🧑 @alandenadel.bsky.social @lcrawford.bsky.social & co

Can we find epistasis in human traits? To help, in our preprint, we present the most scalable and powerful framework for detecting epistasis to date: the “sparse marginal epistasis test” (SME). Thank you @lcrawford.bsky.social, @sampatsmith.bsky.social, Dan Weinreich! doi.org/10.1101/2025... 1/6

Current methods in the field are trained on atlases ranging from 1 to 100 million cells. In our newest preprint, we show that these same approaches tend to plateau in performance with pre-training datasets that are only a fraction of the size.