AlphaGenome Atlas: A predictive map of every possible DNA letter change in the human genome
deepmind ·
Google DeepMind has launched AlphaGenome Atlas, a platform that predicts the molecular effects of all 9 billion possible single-nucleotide variants in the human genome. The resource is built by precomputing predictions from the AlphaGenome AI model, which was previously limited to analyzing specific variants. To aid interpretation, the Atlas introduces the AlphaGenome Variant Impact (AVI) score, combining AlphaGenome and AlphaMissense outputs into a single ranking number applicable to both coding and non-coding regions. The 1-petabyte dataset, which is over 30 times larger than the AlphaFold Database, also includes feature attributions and DNA sequence motifs. Available free for academic research through a website portal, API, and as a Google Antigravity skill, the Atlas has already helped collaborators identify experimentally verified variants in unsolved rare diseases and uncover non-coding associations with common traits. The team views this release as a baseline that will improve as their AI models advance.
deepmind