DeepMind maps 9 billion human DNA variants

Google DeepMind announced AlphaGenome Atlas, described as a predictive map of every possible DNA letter change in the human genome, spanning about 9 billion variants with molecular predictions attached.

The release is a scale-up of AlphaGenome, DeepMind's previously announced sequence-to-function model, rather than a new modeling capability. Its practical value is that predictions are precomputed: researchers screening candidate variants from sequencing studies can query results directly instead of provisioning inference over the genome themselves.

DeepMind credits research collaborators at the University of Exeter, the Broad Institute, Boston Children's Hospital, the Stowers Institute for Medical Research, Harvard University, Memorial Sloan Kettering Cancer Center, the Center for Genomic Medicine at Massachusetts General Hospital, and the University of Kansas Medical Center. Contributors named include Jun Cheng, Kyle R. Taylor, Heidi L. Rehm, Anne O'Donnell-Luria, Julia Zeitlinger, Pushmeet Kohli and Žiga Avsec — a mix of DeepMind staff and clinical-genetics researchers, consistent with a resource aimed at rare-disease and regulatory-variant interpretation.

Details beyond the headline claim are not specified in the material available. The announcement text supplied is limited to acknowledgements, so which molecular outputs the atlas covers, how the predictions were benchmarked, and the access, licensing and download terms are not established here. Researchers should treat the 9-billion figure as the one firm number until the full post and any accompanying paper are reviewed.