DeepMind's AlphaGenome Atlas Uncovers Rare Disease Clues Hidden in Human DNA
By Mashable News Staff
Google DeepMind has launched AlphaGenome Atlas, a free platform offering predictions for 9 billion possible single-nucleotide variants, which is essentially every single-letter DNA change possible in the human genome!
At 1 petabyte, the dataset is over 30 times larger than the AlphaFold Database, and it covers both coding regions (2% of the genome) and non-coding regions (98%) that regulate gene activity.
The platform introduces the AlphaGenome Variant Impact (AVI) score, which merges predictions from AlphaGenome and AlphaMissense into one number, letting researchers rank and interpret variants simultaneously.
Researchers at the Broad Institute used the AVI score to uncover a previously overlooked variant in the DNM1 gene, linking it to epileptic encephalopathy through an abnormal splice site mutation.
A University of Exeter fellow applied the Atlas to genome data from over 54,000 UK Biobank participants and found 22% more non-coding genetic associations tied to protein levels, including genes linked to aging and oxygen sensing.
The same UK Biobank analysis identified 19 genetic regions potentially linked to body mass index by isolating the top 1% of predicted high-impact non-coding variants.
AlphaGenome Atlas is available now for non-commercial academic use via a website portal and API, with commercial access on Google Cloud planned soon, and it can also integrate into Google's Antigravity agentic research system. If this is your area of research, then follow the rabbit trail.