The Information Machine
Updated today·New·first covered 9 Sep 2026·3 sources

Google DeepMind's AlphaGenome Atlas

The gist

Google DeepMind Launches AlphaGenome Atlas to Map 9 Billion DNA Variants

The platform makes AI-predicted effects for every possible single-letter DNA change in the human genome searchable and freely accessible to researchers. The AVI score's coverage of both coding and non-coding regions enabled collaborators to find disease-linked variants and genetic associations that conventional methods missed.

The full picture

AlphaGenome Atlas, released by Google DeepMind on September 8, 2026, is a free academic platform containing AI-predicted molecular effects for all 9 billion possible single-nucleotide variants in the human genome. The dataset is 1 petabyte, more than 30 times larger than the AlphaFold Database. The platform introduces the AlphaGenome Variant Impact (AVI) score, which combines outputs from AlphaGenome and AlphaMissense into a single pathogenicity ranking spanning both protein-coding and non-coding genomic regions. DeepMind states the AVI score provides "best-in-class performance across many variant pathogenicity and rare disease benchmarks." External collaborators applied the score to identify a previously overlooked variant in the DNM1 gene associated with epileptic encephalopathy, with experimental validation confirming the result. Applying Atlas to UK Biobank whole-genome data from over 54,000 participants found 22% more non-coding genetic associations than conventional methods; focusing on the 1% of non-coding variants Atlas rated most impactful, researchers identified 19 genetic regions. Academic access is free through a web portal and API; commercial access through Google Cloud is planned.

How it developed
9 September 2026

Google DeepMind launched AlphaGenome Atlas on September 8, a database of AI-predicted molecular effects for all 9 billion possible single-nucleotide variants in the human genome.

The 1-petabyte dataset, more than 30 times larger than the AlphaFold Database, introduces an AVI score combining AlphaGenome and AlphaMissense to rank variants by pathogenicity across coding and non-coding regions. Collaborators used the AVI score to identify a previously overlooked DNM1 gene variant linked to epileptic encephalopathy, confirmed experimentally, and Atlas found 22% more non-coding genetic associations in UK Biobank data from over 54,000 participants.

8 September 2026

Google DeepMind released AlphaGenome Atlas, a free academic platform with AI-predicted molecular effects for all 9 billion possible single-nucleotide variants in the human genome.

Sources
The daily email

Want this in your inbox?

I send one email each morning with the stories that moved. If you would rather just read here, that works too.

Subscribe free