DeepMind’s free AI DNA mutation predictions cover 9 billion genome variants
Scientists have spent more than two decades holding the human genome’s “instruction manual” without being able to fully read it. That gap just got noticeably smaller. Google DeepMind has released a free, publicly accessible database of AI DNA mutation predictions covering every possible single-letter change that could occur across the human genome — all 9 billion of them, according to the company and reporting from Fortune.
Key takeaways
- Google DeepMind released the AlphaGenome Atlas, a free research database of AI-generated predictions for all 9 billion possible single-letter human DNA mutations.
- The tool interprets both protein-coding DNA and the much larger, harder-to-read regulatory portion of the genome.
- Pushmeet Kohli, DeepMind’s vice president for research, said this marks the first time a researcher anywhere can access a comprehensive map of human genetic variation “by simply opening a browser.”
- The Atlas is free for non-commercial research starting immediately; commercial access will follow soon through licensing on Google Cloud, and sister company Isomorphic Labs will also need that license.
- A paper describing how the Atlas was built has been posted as a preprint on bioRxiv.
Google DeepMind launches AlphaGenome Atlas with AI-driven mutation predictions
The AlphaGenome Atlas is best described as a giant, precomputed catalogue: for every possible swap of one DNA base for another, it estimates what that single change is likely to do to the biological machinery that turns genes on and off. Instead of researchers having to run predictions one variant at a time, the answers are already sitting in the database, ready to be looked up.
… Continue reading the full article at the original source below.
