DeepMind releases AlphaGenome 2 to predict genetic effects in specific cell types
Google DeepMind announced AlphaGenome 2 on November 11, extending its DNA prediction work from broad tissue patterns to particular cell types and biological states. The model is available through the AlphaGenome API for noncommercial research, with commercial access in private preview. DeepMind announcement
The distinction matters because a genetic change can have different consequences in different cells. DeepMind says its redesigned, conditional model combines a DNA sequence with information about the biological condition being studied. Training included more than 107 million individual cell profiles. On the TenK10K evaluation, the company reports identifying 30% of causal variants, compared with 16% for its specified earlier-model baseline. These are developer-reported research results, not evidence of improved patient outcomes. DeepMind announcement
There is also a practical software component to the release. The published documentation separates AlphaGenome 2 from the legacy model, provides a migration guide, and lists tools for encoding cells, selecting supported conditions, predicting variants and scoring them. That gives prospective users a concrete integration path rather than just a headline benchmark. The documentation is mutable and does not establish the release hour. AlphaGenome documentation
For research teams, the opportunity is to prioritize variants and experiments with more relevant cellular context. The next test is whether those predictions remain useful on independent datasets and in follow-up laboratory work. That is an editorial assessment, not a demonstrated outcome of the announcement. DeepMind explicitly says the models are for research and have not been validated or approved for clinical or diagnostic use. Research-use limitation
Sources and citations
Revision history
Published .