Stanford-led AutoScreen is an AI co-scientist that automates CRISPR target discovery, validated on 320+ wet-lab screens
arjunrajlab · x · 2026-09-26
A Stanford/Princeton collaboration (Le Cong, Mengdi Wang, Jure Leskovec, Russ Altman, Aviv Regev et al.) published AutoScreen on bioRxiv, a multi-agent "AI co-scientist" for target discovery in functional genomics.
- Automates the design and analysis of CRISPR screens, closing the loop from AI predictions to real experiments
- Benchmarked on 320+ wet-lab screens
- Ships an "AutoScreen AI hub" with AI-extracted insights from 500+ datasets across BioGrid CRISPR, EMBL expression atlas, and UK Biobank GWAS
- Experimentally validated 3 novel genes that help cancer cells evade immune attack, using fresh unpublished CRISPR screens in melanoma, leukemia, and colon cancer models
- Key members now build scientific AI systems at Phylo, Inc.
Related event: Stanford's AutoScreen Automates CRISPR Target Discovery(2 posts)→
More from Research
- MIT CSAIL's JAZ: minimalist agent with one 'invoke' primitive beats MemGPT at half the cost — omarsar0 · 2026-09-26
- Michael Levin team preprints a pipeline to compensate ion channel mutations with drugs — drmichaellevin · 2026-09-26
- Fact-Checking the NeurIPS-Pangram Story: v3.3.2 Flagged 14.9% of Human-Written Reviews as AI — tughanbulut · 2026-09-26
- StarCraft AI agent Pluto called a 'micro-cheating agent' as human pros still out-micro it — Kangwook_Lee · 2026-09-26
- Fact-Checking "AI Agents Invent Their Own Language": Logs Show Jargon Drift, Not a Secret Code — KhaliSollis · 2026-09-26
- RNA language model trained on sequence alone uncovers new human riboswitches — IgorCarron · 2026-09-26