AlphaGenome Atlas Maps Molecular Effects of All 9 Billion Possible Human DNA Variants
核心洞察
Google DeepMind (搜索) launched AlphaGenome Atlas (搜索), a 1-petabyte platform containing molecular effect predictions for all 9 billion possible single-nucleotide variants in the human genome.
The accompanying AlphaGenome Variant Impact (AVI) score combines AlphaGenome and AlphaMissense predictions into a single number for rapid variant ranking across both coding and non-coding regions.
Collaborators used the Atlas to identify a DNM1 (搜索) variant linked to epileptic encephalopathy (搜索) and to uncover 22% more non-coding genetic associations in over 54,000 UK Biobank participants.
Google DeepMind (搜索) has introduced AlphaGenome Atlas (搜索), a platform containing predictions for the effects of 9 billion single-nucleotide variants — every single-letter change possible — in the human genome. Described as the most comprehensive catalogue of how genetic mutations affect molecular biology, the resource is available for academic research through an intuitive, free-to-use website portal.
The launch addresses a fundamental bottleneck in genomics: with roughly 9 billion possible single-letter mutations in the human genome, testing each one in the laboratory is practically impossible. AlphaGenome Atlas (搜索) builds on AlphaGenome, DeepMind's artificial intelligence model that predicts how genetic variants impact biological processes, by precomputing its predictions at scale to give researchers a genome-wide view of variant effects.
A 1-Petabyte Predictive Map
AlphaGenome Atlas (搜索) is a massive 1-petabyte dataset, more than 30 times larger than the AlphaFold Database. The platform provides several interconnected resources, including molecular effect predictions spanning hundreds of human and mouse cell types and tissues, a collection of over 2,500 recurrent DNA sequence motifs, and the AlphaGenome Variant Impact (AVI) score.
The AVI score combines the strengths of AlphaGenome and AlphaMissense — DeepMind's model for predicting the impact of protein-altering DNA variants — condensing both models' predictions into a single number. This allows researchers to rapidly rank variants and interpret their molecular effects simultaneously. Crucially, the AVI score works for both coding regions (the 2% of the genome that codes for proteins) and non-coding regions (the remaining 98%), which orchestrate gene activity and house most trait-associated variants.
DeepMind reports that the AVI score provides best-in-class performance across many variant pathogenicity and rare disease benchmarks. Each AVI score is also linked to distinct biological features driving it, such as aspects of gene regulation predicted by AlphaGenome or the protein impact score from AlphaMissense, helping researchers interpret which molecular processes — like RNA splicing or gene expression — are predicted to be most disrupted.
Real-World Impact in Rare Disease Research
AlphaGenome Atlas (搜索) has already been applied to targeted research questions by academic partners. In collaboration with the GREGoR Consortium, researchers applied the AVI score to prioritize needle-in-a-haystack genetic variants for unsolved rare disease research. Laura Covill and Anne O'Donnell-Luria from the Broad Institute and their colleagues used the AVI score to prioritize variants that had been overlooked in previous research, discovering a variant affecting a gene called DNM1 (搜索), which is strongly linked to epileptic encephalopathy (搜索).
The AlphaGenome predictions underlying the AVI score showed exactly how the variant functioned: it created an incorrect splice site that led to an abnormal extension of the resulting protein. Experimental screens validated the prediction and found nearby variants with similar effects.
Uncovering Non-Coding Associations in Population Genetics
Beyond individual rare disease research, AlphaGenome Atlas (搜索) can help uncover the genetic architecture of common traits. Gareth Hawkes, a Medical Research Council fellow at the University of Exeter, applied AlphaGenome Atlas to whole-genome data from over 54,000 UK Biobank participants. By grouping rare variants based on their predicted molecular effects, Hawkes uncovered 22% more non-coding genetic associations that would otherwise have been undetectable in statistical noise.
This approach pinpointed specific regulatory variants driving the abundance of critical circulating proteins, including PLA2G7 (搜索) (linked to aging) and EGLN1 (搜索) (a vital cellular oxygen sensor). Extending the approach to body mass index, Hawkes focused on the 1% of non-coding variants that Atlas predicts to be most impactful and identified 19 genetic regions, which could help direct the next stage of targeted research into this trait.
Identifying the Regulatory 'Words' of the Genome
Atlas can also identify which recurring short sequences, or motifs, drive different molecular processes in different cell types. Julia Zeitlinger and Melanie Weilert at the Stowers Institute for Medical Research used this resource to categorize which transcription factors only affect the accessibility of DNA versus which ones are also able to turn genes on and off.
Availability and Future Direction
AlphaGenome Atlas (搜索) is available through a website portal, the AlphaGenome API, and as a skill in Google Antigravity. DeepMind has made the resource accessible for non-commercial use from launch, with commercial use on Google Cloud planned. The AlphaGenome base model is already available for academic use on GitHub and via the AlphaGenome API, and for commercial use on Cloud via Model Garden.
DeepMind views the Atlas as a baseline rather than an endpoint, noting that as its AI models improve, maps of the entire human genome will become increasingly comprehensive and precise. The company positions the resource as a step toward enabling researchers and industry partners to find novel therapeutic targets, better understand genetic disorders, and drive the next wave of targeted experimental validation.