Genome atlas reportedly opens predictions for 9 billion DNA-letter changes

Searchable predictions for nine billion variants

Nature describes Google DeepMind’s AlphaGenome Atlas as an AI-generated resource that assigns predicted effects to all nine billion possible single-letter variants in the human genome.

The atlas is reported to be freely available for non-commercial use and to include more than 100 million short insertions and deletions observed in human genomes. It also provides an AlphaGenome Variant Impact (AVI) score intended to summarize a variant’s predicted biological effects.

Researchers could access AlphaGenome predictions through an API, but doing so required writing code, according to DeepMind. Applying models of this scale across the whole genome was also described as infeasible for many researchers.

DeepMind has reportedly made a non-commercial AlphaGenome Atlas available, providing model-generated predictions for every possible single-letter change in the human genome and for more than 100 million observed short insertions or deletions.

The atlas reportedly widens access to predictions without requiring code, while model outputs remain distinct from experiments. Researchers quoted in the article caution that the resource does not replace experimental work or individual clinical assessment.