Google DeepMind has released AlphaGenome Atlas, a 1-petabyte dataset containing predicted molecular effects for all approximately 9 billion possible single-letter DNA changes in the human genome2. The company described it as the most comprehensive catalogue of how genetic mutations affect molecular biology.
AlphaGenome Atlas contains what the researchers called a "predictive map of every possible DNA letter change in the human genome"1. DNA is encoded in four chemical letters — A, C, G, and T — and the human genome contains roughly three billion letter pairs. The dataset covers the predicted molecular consequences of substituting any one of those letters at every position across the genome.
Google DeepMind scientists said the platform could help unravel mysteries of the human genome and transform understanding of biology, accelerating scientific research and ultimately paving the way for new disease treatments. The researchers published a blog post detailing the release on Tuesday.
ANALYSIS At 1 petabyte, the dataset's scale reflects the combinatorial scope of the problem: approximately 9 billion possible single-nucleotide variants, each annotated with predicted molecular effects. Making these precomputed predictions publicly available could reduce the computational burden on individual research groups that would otherwise need to run variant-effect models themselves. The release extends Google DeepMind's genomics line, which previously included AlphaFold for protein-structure prediction, into the domain of variant interpretation at whole-genome scale.