AlphaGenome Atlas: a high-resolution map of human DNA
Points and comments are a snapshot, not live.
Google DeepMind releases AlphaGenome Atlas, predicting all 9 billion possible single-letter DNA changes.
The AlphaGenome Atlas by DeepMind predicts the regulatory impact of every possible single nucleotide variant in the human genome, pre-computed for all 9 billion single-letter changes into a 1-petabyte dataset. It introduces the AlphaGenome Variant Impact (AVI) score for both coding and non-coding regions, accessible via a no-code web portal. Early applications include identifying a critical variant in the DNM1 gene for rare disease research and uncovering 22% more non-coding genetic associations linked to BMI from UK Biobank data.
What commenters are saying
Commenters split between appreciation for the resource and skepticism about utility and commercial intent. Some questioned whether the predictions are clinically meaningful, noting that individual SNP predictions often lack pathogenicity context. Others pointed out that 23andMe data cannot fully leverage this tool due to limited sequencing. Several highlighted the Terms of Service restricting commercial use, with speculation that DeepMind may sell access via Isomorphic Labs. A key counterpoint came from a critique referencing Katie Pollard's ISMB talk: existing human variation alone may be insufficient for accurate variant prediction, requiring cross-species data.