DeepMind's AlphaGenome 2 identifies 30% of causal DNA variants, up from 16%
Spirited-Sir-3034 · reddit · 2026-11-12
Google DeepMind introduced AlphaGenome 2, a DNA model that predicts how genetic variants affect specific cell types and cellular states rather than relying on broad tissue-level averages.
- Trained on over 107 million human and mouse cell profiles.
- Identified 30% of causal variants on the TenK10K immune-cell benchmark, up from 16% for its predecessor and other specialized tools.
- Predicted whether variants increase or decrease gene expression in 43% of cases, up from 23%.
- Could help researchers prioritize disease-linked variants for lab study, but predictions still require experimental validation; not approved for clinical diagnosis.
More from AGI Musings
- Academia is developing an 'AI caste system,' warns investor as models erase research moats — sebkrier · 2026-11-12
- Gary Marcus endorses 2031-2036 as a more reasonable AGI timeline — GaryMarcus · 2026-11-12
- Philosopher Yuk Hui's new book rewrites Kant's three questions for machines — tarantulae · 2026-11-12
- SpaceX's Southaven data centers net local schools and county $63M a year — 0xAllen_ · 2026-11-12
- New Yorker Cartoon: It All Began With Letting Autocomplete Finish Our Sentences — miniapeur · 2026-11-12
- AI Developer Burnout: 10x Speed Just Means Unmanageable Ticket Backlogs — No-Common1466 · 2026-11-12