Google DeepMind released AlphaGenome Atlas, a deep learning system that provides precomputed predictions for all 9 billion possible single-nucleotide variants in the human genome. The system combines molecular effect predictions across multiple cell types with a unified AlphaGenome Variant Impact (AVI) score that integrates both gene-regulation and protein-impact assessments. It addresses the practical impossibility of lab-testing every possible mutation, making predictions accessible through a web portal, an API, and integration with Google Antigravity for research workflows.
