1/ How to keep, and profit from, British Antarctica It's eight times the size of Great Britain and is rich in resources – oil and gas, plus (judging by the geology) plenty of rare earths. The jewel is the peninsula: comparatively mineable, but already contested. A proposal🧵

The discussion around virtual cell models has primarily focused on algorithms and metrics. What about the training data? Is scRNA-seq good enough? A key challenge is distinguishing direct vs indirect effects on gene expression. 1/5

💡 Apparently, if you throw 10+ proteins into each AlphaFold3 run, you get whole interactomes on the house😂🧬 “Pooled” folding reduces # jobs 100×, halves runtime, and even spots novel PPIs

🧠🔬 A new microscope, ExA-SPIM, lets researchers zoom from molecules to whole brains, no slicing required. By combining tissue expansion with high-speed imaging, it captures stunning detail across entire mouse brains and even human samples. elifesciences.org/articles/91979…

Excited to share a milestone published in Nature Medicine from our decade-long effort to build The Human Phenotype Project, a unique longitudinal cohort with unmatched depth of clinical and multi-omic profiling, enabling truly predictive, personalized medicine. Led together with

While UK Biobank enabled access to population-based proteomics at scale, most omics studies in disease-focused cohorts still suffer from small sample sizes. The Global Neurodegeneration Proteomics Consortium brought together 35,000 serum, plasma, and CSF samples from

IMO, medium scale combinatorial regulator perturb expts with diverse measurable stoichiometry are going to be far more informative to learn causal models of cis-trans regulation than genome scale single gene knockouts. Hope we get more of these kinds of datasets at scale. 1/

I'm excited to share work on a research direction my team has been advancing: connecting machine learning derived genetic variant embeddings to downstream tasks in human genetics. This work was led by the amazing Divyanshi Srivastava! biorxiv.org/content/10.110…

Probably the most promising use of proteomics in drug development is creating signatures that act as surrogates for hard endpoints in early phase trials. In a MASH trial, semaglutide showed dose-dependent effects on serum proteomic signatures of liver histology mirroring actual

Enriching polygenic risk scores with single-cell ATAC-seq about chromatin accessibility in disease-relevant cell types shows potential to 👉improve predictive performance 👉better understand cells driving genetic predisposition to disease 👉uncover specific molecular drivers

Excited to share a review paper we have written on statistical fine-mapping!

Thanks to Riya Sinha in my lab, IGV Team browser will natively support dynseq (dynamic sequence tracks) in an upcoming release. These tracks are very useful to directly visualize base-resolution scores (e.g. contribution scores from ML models, conservation etc). 1/

Today the British government sent a clear message: massacre innocent civilians and get rewarded. Recognising Palestine doesn’t bring peace or alleviate suffering - it rewards terror, legitimises barbarism and emboldens our enemies everywhere.

The City Bridge Foundation is an ancient fund with deep pockets whose specific purpose is to build and maintain London’s bridges. It could afford to fix Hammersmith bridge & build the Garden Bridge, So why is it paying for woke charities and Nimbys instead?

Running back calculations on this, the one off stock gift to ARM employees would now have been worth $22B Blocking the NVIDIA deal and missing out on Britain being ~15% of the first $4T company was *the* worst recent tech industrial strategy mess up. Now (of course) forgotten.

"BBNs" have officially been given a government name: Frontier Research Contractors. The reason why: ARIA has funded a program to help found and scale them! This is a testament to ARIA's ambition. It's the only pot of funds in the world dedicated to launching these orgs

Stimulating perspective on neutrophil heterogeneity and function. A must-read from #ThePowerOfMany

A new machine learning tool, ProDomino, can predict where two proteins are able to be “fused” together. This seems really useful, in part, because many tools in biotechnology are made by fusing protein domains. So this tool, in principle, will speed up tool creation. Base

Our paper is now out in final form at Nature Genetics! For those who missed the preprint, we used large-scale Perturb-seq targeting transcription factors to push primary fibroblasts into diverse transcriptional states, including those observed in cell atlas studies.

I hate reference limits of journals so much. I'm sitting here trying to find a few more references to remove despite me knowing that I'll either remove context & evidence for scientific conclusions or not mention tools that very much deserve to be mentioned and cited.