Google DeepMind unveils AlphaGenome 2 DNA variant model
According to Google DeepMind via Demis Hassabis and the AlphaGenome docs site, AlphaGenome 2 predicts cell type specific impacts of DNA variants and is available for academic research through an API.
SourceAnalysis
Google DeepMind announced AlphaGenome 2 on November 11 2026, introducing a new DNA sequence model that predicts how variants affect specific cell types such as those in the heart, brain, and immune system.
The model addresses the challenge of a shared DNA sequence across tissues by focusing on cell-specific impacts, providing researchers with tools to study drivers of complex diseases. It is available for academic research through an API, according to the announcement from Demis Hassabis and Google DeepMind.
Practical Applications in Genomics
This development allows scientists to examine variant effects at a cellular level, supporting more targeted investigations into disease mechanisms without requiring broad assumptions across all tissues.
Workflow Impact for Researchers
- Enables cell-type specific predictions that refine experimental design in genomics labs.
- Integrates with existing academic pipelines via the provided API access.
FAQ
What does AlphaGenome 2 predict?
It predicts the impact of DNA variants within particular cell types to aid understanding of complex diseases.
Who can access the model?
Academic researchers can use it through the API as stated in the announcement.
Demis Hassabis
@demishassabisNobel Laureate and DeepMind CEO pursuing AGI development while transforming drug discovery at Isomorphic Labs.