DeepMind's AlphaGenome model and AlphaGenome Atlas provide genome-wide variant impact predictions
Google DeepMind's AlphaGenome is a predictive model that analyzes up to 1 million base-pair sequences and was used to generate the AlphaGenome Atlas, a dataset of predicted effects for all ~9 billion single-nucleotide changes; the Atlas provides a single AVI (AlphaGenome Variant Impact) score to help researchers prioritize variants. DeepMind launched AlphaGenome in June 2025 and introduced AlphaGenome Atlas in September 2025, and early users at institutions like the University of Exeter and the Broad Institute report using AVI to narrow candidate variants in research.