Google DeepMind has released "AlphaGenome Atlas," a database that predicts the effects of approximately 9 billion single nucleotide substitutions (single-letter changes in DNA bases) in the human genome. This platform uses the AI model "AlphaGenome" to pre-calculate the impact of genetic changes in the human genome on regulatory functions.

The dataset spans approximately 1 petabyte and covers extensive information, including not only protein-coding regions but also non-coding regions (about 98% of the genome) that control gene expression. By using a metric called the "AlphaGenome Variant Impact (AVI) score," researchers can rapidly evaluate and prioritize the effects of genetic mutations on molecular biological processes. This score integrates predictions from two models: AlphaGenome and AlphaMissense.

AlphaGenome Atlas is available for non-profit academic research via a web portal. In addition to API access, commercial availability through Google Cloud is expected to be provided soon. Validations by external collaborators have reported that the tool has contributed to identifying causal mutations for rare diseases and discovering non-coding regions associated with common traits.

Source: