Google DeepMind launched AlphaGenome Atlas today, a free 1-petabyte database that precomputes the predicted effect of all 9 billion possible single-letter mutations across the human genome, condensed into one AlphaGenome Variant Impact score researchers can look up with no code.
Read the full story: Google DeepMind Maps All 9 Billion Human DNA Mutations →
Transcript
Google DeepMind just switched on AlphaGenome Atlas, and it already has an answer for every possible mutation in the human genome. All nine billion of them. Not run one at a time. Precomputed, searchable in seconds. Each variant gets a single score, the AlphaGenome Variant Impact score, that rolls up eleven separate predictions, gene expression, chromatin, splicing, into one number a researcher can sort by.
It's free for non-commercial research through a no-code web portal, plus an API and a skill inside DeepMind's own coding platform. The dataset is about a petabyte, more than thirty times the size of the AlphaFold Database. Outside experts call it the field's leading model, but warn the score is easy to misread. The real test starts now: does the wet lab confirm what the model predicted.