Google DeepMind Releases AlphaGenome Atlas (blog.google)

Google DeepMind has launched AlphaGenome Atlas, a groundbreaking database providing a high-resolution map of the human genome, focusing on the 98% of non-coding DNA that remains poorly understood. This innovative resource utilizes the AlphaGenome AI model to predict the effects of all 9 billion possible single nucleotide variants, resulting in a massive 1-petabyte dataset. Central to this release is the AlphaGenome Variant Impact (AVI) score, a simplified metric that combines predictions for both coding and non-coding regions, enabling researchers to efficiently prioritize variants without wading through extensive data. The significance of AlphaGenome Atlas lies in its potential to empower the scientific community by accelerating research into rare genomic variations and complex traits. For instance, researchers at the Broad Institute leveraged the AVI score to identify critical genetic variants involved in rare diseases, while Dr. Gareth Hawkes used it to uncover more associations between non-coding variants and body mass index in a large cohort from the UK Biobank. Importantly, the Atlas is accessible through a user-friendly website, democratizing genomic research for clinical researchers and biologists globally and fostering a new era of biological discovery.