DeepMind Launches AlphaGenome Atlas for 9 Billion DNA Variants
Single-nucleotide changes across the human genome can alter molecular processes and influence disease risk, but testing each variant experimentally is costly and slow. Google DeepMind’s AlphaGenome uses artificial intelligence to predict the molecular effects of DNA variants, helping researchers prioritize candidates for laboratory study. The atlas could accelerate work on disease mechanisms and point scientists toward potential targets for new treatments.
Google DeepMind has released the AlphaGenome Atlas with precomputed molecular-effect predictions and AVI scores for about 9 billion human single-nucleotide variants. Generated through inference by the AlphaGenome AI model, the dataset totals about 1 petabyte. Researchers can now query the results through an academic web portal and an API, avoiding the need to run the model separately for every variant under investigation.
All Coverage
2 original reportsThe Backstory
The history behind this eventGoogle DeepMind Disbands AlphaFold Team in Broader Science AI Shift
AlphaFold established Google DeepMind as a leader in AI-driven science after the system largely solved the 50-year protein-folding challenge in 2020. Its database now covers more than 200 million protein structures and is used by over 3 million researchers. The achievement earned DeepMind Chief Executive Demis Hassabis and researcher John Jumper a share of the 2024 Nobel Prize in Chemistry, demonstrating how specialized AI could reshape biology and drug discovery.
Google DeepMind has eliminated AlphaFold as a standalone team, according to reports published July 29, as it shifts from tackling individual scientific problems toward broadly capable, Gemini-powered discovery systems. Researchers have been reassigned to Gemini, AI coding, genomics, enzyme design, nuclear fusion and Alphabet-backed Isomorphic Labs. Jumper, Jonas Adler and Alexander Pritzel have left for Anthropic, while nearly a quarter of the authors of the original AlphaFold paper have departed DeepMind. The company said its evolving structure is intended to apply AI across a wider range of scientific fields.
Subscribe to Mark Radar Weekly
Every Friday, the week's strongest signals in your inbox. Unsubscribe anytime.
If you search news on Google, you can set Mark Radar as a preferred source—our coverage will show up more often in your results. Set as preferred source on Google →