AlphaGenome Atlas

AlphaGenome Atlas is a database introduced by Google DeepMind that predicts the effects of every possible single nucleotide variant in the human genome. It was created by using the AlphaGenome AI model to pre-calculate the regulatory impact of all nine billion single-letter genetic changes, producing a massive, one-petabyte dataset. The Atlas is intended for researchers, clinical researchers, and biologists, who can explore it through an intuitive website portal that requires zero coding skills, with API and Antigravity access available for deeper research. Its stated purpose is to provide grounded genomic insights that will accelerate the pace of biological discovery. The human genome is made of about three billion base pairs of DNA, but much of it remains a mystery. Scientists understand the roughly two percent of the genome that codes for proteins relatively well, yet they have only limited knowledge of the remaining ninety-eight percent. Google DeepMind's AlphaGenome model had already shown how single changes in these non-coding DNA regions can disrupt molecular processes such as protein production, but the bigger picture across the genome remained unclear. At the same time, identifying rare, non-coding variants linked to complex traits is difficult because of statistical noise in the data. AlphaGenome Atlas was created to address this gap by turning an enormous space of possible mutations into a searchable, precomputed resource that researchers can rapidly query instead of evaluating variants one at a time. A central feature of the Atlas is the AlphaGenome Variant Impact (AVI) score. This single, easy-to-use score combines predictions for both coding and non-coding regions of the genome, giving researchers one measure to work with instead of a large collection of separate outputs. Because the score condenses coding and non-coding predictions into a single number, it allows researchers to quickly prioritize the most promising avenues for research without sifting through thousands of data points. The AVI score is described as the mechanism that helps scientists rapidly navigate the vast information held in the Atlas, making it possible to rank variants by predicted impact and focus attention where it is most likely to matter. In practice, this scoring approach is what turns a one-petabyte dataset into something a research team can act on. The Atlas covers every possible single nucleotide variant in the human genome. Google DeepMind used the AlphaGenome AI model to pre-calculate the regulatory impact of all nine billion single-letter genetic changes, resulting in a one-petabyte dataset. Crucially, these predictions span both coding and non-coding regions, rather than focusing only on the small fraction of the genome that codes for proteins. This matters because the non-coding portion of the genome is where much of the remaining biological mystery lies, and where single-letter changes can disrupt molecular processes like protein production. By precomputing predictions across the full set of possible variants, the Atlas removes the need for researchers to run predictions on demand for each variant they want to investigate, and instead gives them a comprehensive catalogue of how genetic mutations affect molecular biology. Access to the Atlas is designed to be broad. It is available through an intuitive website portal that requires zero coding skills, which the announcement frames as democratizing access for clinical researchers and biologists worldwide. For researchers who need to work at greater depth, the Atlas also offers API access, along with access through Antigravity. AlphaGenome Atlas is free to explore through the visual web interface. This combination of a no-code portal and programmatic access means the same underlying predictions can serve a biologist inspecting a single variant in a browser and a research group building variant-prioritization workflows into its own pipelines. The Atlas works by precomputation rather than on-demand prediction. Rather than running the AlphaGenome model each time a researcher wants to understand a variant, Google DeepMind used the model to pre-calculate the regulatory impact of every possible single-letter genetic change in the human genome — all nine billion of them — and stored the results as a one-petabyte dataset. The Atlas then helps scientists rapidly query this vast information, and it introduces the AVI score as a way to summarize the underlying predictions into a single usable ranking. This approach shifts the heavy computational work to a one-time, genome-wide precomputation and turns the result into something that can be explored interactively or accessed programmatically, which is what makes rapid navigation of the full variant space feasible. The stated benefit of AlphaGenome Atlas is that it provides grounded genomic insights that will accelerate the pace of biological discovery. By combining coding and non-coding predictions into a single AVI score, it lets researchers prioritize the most promising avenues for research without sifting through thousands of data points, saving effort that would otherwise be spent on manual triage. The no-code web portal broadens who can use these predictions, democratizing access for clinical researchers and biologists worldwide rather than limiting it to those who can write code or run models themselves. The announcement also describes the Atlas as acting as a powerful augmentation partner for the scientific community, accelerating research — a claim illustrated by the rare disease and complex trait examples it cites. Overall, the Atlas is positioned as a resource that makes a previously unclear genome-wide picture accessible and actionable. AlphaGenome Atlas is already being applied in real research settings. At the Broad Institute, Laura Covill and her team used the AVI score to prioritize variants for unsolved rare disease research; the tool highlighted a critical variant in the DNM1 gene, predicting that it created an incorrect splice site, which provided crucial supporting evidence that helped successfully solve the case. In a second example, Dr. Gareth Hawkes applied AlphaGenome Atlas to data from more than 54,000 UK Biobank participants to study complex traits. By grouping variants based on predicted molecular effects, he uncovered 22% more non-coding genetic associations, and by focusing on the top 1% of impactful variants he identified 19 genetic regions linked to body mass index (BMI), directing the next stage of targeted research. These examples show the Atlas being used both to prioritize individual candidate variants in rare disease and to group and rank variants across large population cohorts. AlphaGenome Atlas is aimed at the scientific community — researchers, clinical researchers, and biologists — including teams working on rare genomic variation and on complex traits. It is described as available today through an intuitive website portal that requires zero coding skills, with API and Antigravity access for deeper research, and it is free to explore. The Atlas is presented as part of Google DeepMind's ongoing commitment to accelerate genomic discovery and science, for everyone, and it builds on the earlier AlphaGenome model. For those who want more detail, the announcement points readers to the Google DeepMind blog. AlphaGenome Atlas is Google DeepMind's high-resolution, precomputed map of how single-letter changes in human DNA affect molecular biology. By calculating predictions for all nine billion possible single nucleotide variants and condensing them into the AlphaGenome Variant Impact score, it gives researchers a practical way to explore both coding and non-coding regions of the genome and prioritize the variants most worth pursuing. Free, accessible through a no-code portal, and supported by API and Antigravity access, it aims to serve as an augmentation partner for the scientific community and to accelerate the pace of biological discovery.