Google DeepMind Maps Predicted Effects of 9 Billion Human DNA Variants

Google DeepMind has introduced AlphaGenome Atlas, an artificial intelligence-powered resource containing predictions for the effects of more than 9 billion possible single-letter changes in the human genome.

Google DeepMind's AlphaGenome Atlas provides predicted molecular effects for more than 9 billion possible single-letter changes in the human genome. [Photo: Google DeepMind]

DeepMind said the Atlas is a precomputed catalogue of predicted molecular effects for every possible single-nucleotide variant, or single-letter DNA change.

The resource is designed to help researchers investigate how genetic variations may affect processes such as gene expression, RNA splicing and chromatin accessibility.

The Atlas builds on DeepMind's AlphaGenome model, which predicts how genetic variants can affect biological processes.

Rather than requiring researchers to run the model separately for individual variants, the Atlas precomputes predictions across the genome.

DeepMind said the dataset is about 1 petabyte in size and covers more than 9 billion possible single-letter changes.

The company has made the resource available for academic and other non-commercial research through a free online portal. Commercial access is planned through Google Cloud.

The resource also includes an AlphaGenome Variant Impact score designed to help researchers rank variants according to their predicted biological effects.

Pushmeet Kohli, DeepMind's vice president of research and head of its AI for science team, has described the project as part of a broader effort to make the information contained in the human genome more understandable.

The human genome contains about 3 billion DNA base pairs, meaning that there are more than 9 billion possible single-letter substitutions.

Testing each change experimentally would be impractical, making computational predictions potentially valuable for prioritizing variants for further research. However, AlphaGenome Atlas provides predictions, not laboratory confirmation of the biological effects of every mutation.

DeepMind itself describes the Atlas as a baseline and says the predictions will become more comprehensive and precise as its models improve.