On September 8, 2026, Google DeepMind announced the launch of AlphaGenome Atlas, a platform that provides predictions for the molecular effects of 9 billion single-nucleotide variants—every possible single-letter change in the human genome. This resource aims to accelerate biological research by offering the most comprehensive catalogue of how genetic mutations impact molecular biology.
The platform builds on AlphaGenome, an AI model previously developed by DeepMind to predict variant effects. By precomputing predictions at scale, AlphaGenome Atlas offers researchers an accessible, big-picture view of variants across the entire genome, similar to how a geographic atlas maps terrain features.
A key feature is the AlphaGenome Variant Impact (AVI) score, which combines predictions from AlphaGenome and AlphaMissense into a single number, enabling rapid ranking and interpretation of variants. The AVI score works for both coding and non-coding regions, covering the 98% of the genome that regulates gene activity.
The dataset is massive—1 petabyte, over 30 times larger than the AlphaFold Database—and includes thousands of molecular effect predictions per variant, AVI feature attributions, and over 2,500 DNA sequence motifs. External collaborators have already used it to identify and experimentally verify variants in rare disease research.
AlphaGenome Atlas is available through a free website portal, an API, and as a skill in Google Antigravity, making it accessible to researchers without coding expertise.