AlphaGenome Atlas: A predictive map of every possible DNA letter change in the human genome

Google DeepMind has launched AlphaGenome Atlas, a platform offering predictions for every single-nucleotide variant in the human genome—approximately 9 billion possible single-letter DNA changes. The resource is designed to help researchers interpret how genetic mutations affect molecular biology, addressing a longstanding challenge in genomics. By precomputing these predictions at scale, the platform provides a comprehensive view of variant impacts across the entire genome, enabling faster and more systematic analysis than traditional lab-based methods.

The platform introduces the AlphaGenome Variant Impact (AVI) score, which combines predictions from AlphaGenome and AlphaMissense to rank variants by their potential molecular effects. Unlike previous tools, the AVI score works for both coding and non-coding regions of the genome, covering 2% and 98% respectively. Researchers can use the score to quickly identify high-impact variants and explore which molecular processes—such as RNA splicing or gene expression—are disrupted, supported by feature attributions that highlight specific mechanisms.

AlphaGenome Atlas has already been applied in rare disease research, where collaborators used the AVI score to prioritize variants in unsolved cases. In one example, researchers identified a variant in the DNM1 gene linked to epileptic encephalopathy by detecting an incorrect splice site that altered protein function. Experimental validation confirmed the prediction, demonstrating the platform's utility in pinpointing causal variants amid genomic noise.

The platform also supports broader genetic studies by helping researchers uncover associations between rare non-coding variants and common traits. In a study of over 54,000 UK Biobank participants, AlphaGenome Atlas increased the detection of non-coding genetic associations by 22%, revealing regulatory variants tied to proteins like PLA2G7 and EGLN1. Additionally, the tool can identify recurring DNA motifs that drive molecular processes, such as transcription factor binding sites, offering deeper insights into gene regulation across different cell types.