DeepMind releases AlphaGenome 2, a model that predicts gene effects by cell type
On November 11, 2026, Google DeepMind introduced AlphaGenome 2, which predicts DNA variant effects for one tissue or cell type at a time. DeepMind says it finds nearly twice as many expression-changing variants as its prior best model.
What changed
AlphaGenome is now a conditional model. On the TenK10K dataset DeepMind reports 30% vs. 16% causal variants identified and 43% vs. 23% directional gene effects against earlier tools.
Why it matters
Better variant-effect prediction could help researchers narrow down which DNA changes drive disease.
What remains uncertain
The performance numbers are DeepMind’s own, and the post gives two different figures for training data. The tool is research use only, not validated for clinical or diagnostic use.
What to watch
Watch for independent benchmarks and for uptake through the AlphaGenome API.
Source
Why this ranks here
A clear science result from a major lab, but limited to research use for now.
- Impact7
- Reach6
- Novelty7
- Commitment6
- Coverage quality7
- Future potential7