DeepMind ships AlphaGenome Atlas for 9 billion DNA variants

Google DeepMind published AlphaGenome Atlas, a catalogue of precomputed molecular-effect predictions for about 9 billion single-nucleotide variants — which it calls every possible single-letter change in the human genome.

A 9-billion-variant DNA map is now free for academic research — DeepMind says it is not a clinical tool. Partner case notes stay the lab’s.

DeepMind posted AlphaGenome Atlas. The lab describes a catalogue of predicted molecular effects for roughly 9 billion SNVs — single-nucleotide variants, meaning one-letter DNA swaps. In its words, that is every possible single-letter DNA change.

To rank variants, DeepMind releases an AlphaGenome Variant Impact (AVI) score that condenses AlphaGenome and AlphaMissense predictions into one number. It says the score covers coding regions (DNA that makes proteins) and non-coding regions (the rest of the genome).

The dataset is described as about 1 petabyte — a million gigabytes. Non-commercial academic use is free through a website portal and the AlphaGenome API. A skill in Google Antigravity is listed as available now; commercial access on Google Cloud is listed as planned.

DeepMind frames the resource as research, not a clinical tool. Collaborator notes on the page — rare-disease and UK Biobank examples — are DeepMind’s account of partner work.