Google DeepMind launches AlphaGenome Atlas mapping 9 billion DNA variants

Google Alphabet Inc. ↗ 0.02% DeepMind on Tuesday released AlphaGenome Atlas, a database containing predictions for the molecular effects of all 9 billion possible single-letter DNA changes in the human genome — an effort the company describes as "the most comprehensive catalogue of how genetic mutations affect molecular biology."deepmind

The 1-petabyte dataset, more than 30 times the size of DeepMind's Nobel Prize-associated AlphaFold Database, is available for free to academic researchers through a no-code web portal, an API, and as a skill within Google's Antigravity agentic platform. Commercial access on Google Cloud is expected soon.blog

From Model to Map

AlphaGenome Atlas builds on the AlphaGenome model, first published in Nature in January 2026, which predicts how single DNA mutations affect cellular processes such as gene expression and RNA splicing. Rather than requiring researchers to run one query at a time, DeepMind pre-computed predictions for every possible single-nucleotide variant across the human genome. "Basically it took us some time to really precompute and also analyze this many variants because the space is so big," said Žiga Avsec, DeepMind's genomics lead, in a press briefing reported by The Verge.theverge+1

Alongside the raw data, DeepMind introduced the AlphaGenome Variant Impact (AVI) score, which condenses predictions from both AlphaGenome and AlphaMissense into a single number per variant. The score works across the roughly 2 percent of the genome that codes for proteins and the remaining 98 percent of non-coding regions — the "dark matter" where most disease-linked genetic variation resides.scientificamerican+1

Early Results in Rare Disease and Population Genetics

External collaborators have already put the Atlas to use. Working with the GREGoR Consortium, researchers Laura Covill and Anne O'Donnell-Luria at the Broad Institute used the AVI score to identify a variant in the DNM1 gene linked to epileptic encephalopathy. Atlas predicted the variant created a faulty splice site, producing an abnormally extended protein — a finding later confirmed by lab experiments.unite+1

Separately, Gareth Hawkes at the University of Exeter applied Atlas to whole-genome data from more than 54,000 UK Biobank participants and uncovered 22 percent more non-coding genetic associations than standard methods, including variants tied to aging-related protein PLA2G7 and the cellular oxygen sensor EGLN1.deepmind+1

A Familiar Playbook

The release mirrors the strategy DeepMind followed with AlphaFold, which expanded from roughly 190,000 experimentally known protein structures to more than 200 million AI-predicted ones and became one of the most widely used tools in modern biology. "This represents the first time that any researcher in the world can access a comprehensive map of the human genome and its variations by simply opening a browser," said Pushmeet Kohli, vice president of science at Google DeepMind, according to Scientific American.scientificamerican+1

DeepMind cautioned that Atlas is a research tool, not a clinical diagnostic, and noted gaps in the underlying model's training data. Avsec described it as "a baseline rather than an endpoint" that will improve as the models evolve.deepmind+1