Google DeepMind has announced "AlphaGenome Atlas," a platform that predicts the biological impact of approximately 9 billion single nucleotide variants (single-letter changes in DNA) within the human genome.

This platform consists of the prediction results from the "AlphaGenome" AI model, announced by the company last year, pre-calculated and organized as a large-scale catalog covering the entire genome. Atlas provides prediction data for extensive genomic regions, including not only protein-coding regions but also non-coding regions that regulate gene function. The scale of the data reaches approximately one petabyte.

Alongside this, the "AlphaGenome Variant Impact (AVI) score," which quantifies the impact of mutations, has also been released. This score combines predictions from two models, AlphaGenome and AlphaMissense, enabling researchers to rapidly rank the molecular-level effects of variants.

AlphaGenome Atlas is available for free to academic research via a website starting today, with commercial availability expected soon through Google Cloud.

Source: