AlphaGenome is an AI model developed by Google DeepMind for predicting the effects of genetic variants on gene regulation in the human genome. It was published in Nature on January 28, 2026, and continues the pattern, established by the AlphaFold family, of DeepMind applying frontier AI to problems in biology.
Publication
The model was described in "Advancing regulatory variant effect prediction with AlphaGenome," published in Nature on January 28, 2026. The New York Times covered the release the same day under the headline "With AlphaGenome, Researchers Are Using A.I. to Decode the Human Blueprint."
AlphaGenome Atlas
Google DeepMind released AlphaGenome Atlas on September 8, 2026, a dataset of roughly one petabyte holding precomputed predictions for all nine billion possible single-nucleotide variants in the human genome — more than thirty times the size of the AlphaFold Database. It ships with the AlphaGenome Variant Impact score, which condenses AlphaGenome and AlphaMissense predictions into a single number, and a compendium of more than 2,500 recurrent DNA sequence motifs. Gareth Hawkes of the University of Exeter applied it to whole-genome data from over 54,000 UK Biobank participants and reported uncovering 22 per cent more non-coding genetic associations. The dataset is free for non-commercial academic use through a web portal and the AlphaGenome API, with commercial availability on Google Cloud described as forthcoming (Source: deepmind.google).
The Atlas shifts the model's mode of use: rather than querying the model per variant, researchers can consult a precomputed map, which is the same distribution pattern DeepMind used for the AlphaFold Database. See AI for Science.
Related models
AlphaGenome extends a sequence of DeepMind biology models: AlphaFold (AlphaFold (DeepMind)) for protein structure prediction, AlphaFold 3 for molecular complexes, and AlphaGenome for genome regulation. The 2024 Nobel Prize in Chemistry was awarded for AlphaFold, cited as an example of AI applied to scientific problems. Demis Hassabis leads DeepMind's work in this area, and the broader vision of AI-accelerated biology is set out in Machines of Loving Grace (Amodei's biology vision).
The model is relevant to AI Biosecurity as a dual-use concern and to AI for Science.
Relationships
- developer: Google DeepMind
- predecessor-family: AlphaFold (DeepMind)
- related: AI for Science, Demis Hassabis, AI Biosecurity, Machines of Loving Grace (Amodei biology vision).
Sources
- Nature / Advancing regulatory variant effect prediction with AlphaGenome (2026-01-28)
- NYT / With AlphaGenome, Researchers Are Using A.I. to Decode the Human Blueprint (2026-01-28)
- Google DeepMind / AlphaGenome Atlas: A predictive map of every possible DNA letter change in the human genome (2026-09-08) — (Source: deepmind.google)