AlphaGenome Atlas makes nine billion DNA variant predictions searchable
Google DeepMind has released a research atlas of predicted molecular effects for possible single-letter DNA changes. The resource helps researchers prioritize variants for investigation; its predictions still need validation.
Google DeepMind introduced AlphaGenome Atlas on September 8, making precomputed predictions for roughly nine billion possible single-letter changes in human DNA accessible to researchers.
The atlas builds on AlphaGenome's predictions of molecular effects. A companion AlphaGenome Variant Impact score combines signals from AlphaGenome and AlphaMissense to help researchers rank variants that may deserve closer attention.
Access is available through a web portal, the AlphaGenome API and an integration with Google Antigravity. DeepMind describes the portal as free for academic research and says collaborators have already used the resource to identify variants for experimental follow-up.
The useful distinction is between predicted molecular impact and demonstrated biological consequences. This atlas can help scientists choose what to test. A high score is not, by itself, proof that a variant causes disease or a substitute for clinical interpretation.