DeepMind releases AlphaGenome Atlas, precomputed predictions for 9 billion possible DNA variants

Google DeepMind/AlphaGenome

DeepMind published the AlphaGenome Atlas, predictions for every possible single-letter change in the human genome, about 9 billion variants in a petabyte of data. A new AVI score combines AlphaGenome with AlphaMissense to rank variants in coding and non-coding DNA. It is free for academic research.

• About 30 times the size of the AlphaFold Database.
• Partners include the Broad Institute, University of Exeter, Boston Children's Hospital and the GREGoR Consortium.
• Found 22% more non-coding associations in UK Biobank data, per DeepMind.
• Commercial Atlas access is due on Google Cloud, which already offers the base model; not validated for clinical use.