GoKawiil
Tech News
Google DeepMind has released AlphaGenome 2, a successor to its AlphaGenome genome-prediction model, trained on more than 100 million individual cell profiles across 68 tissues. The model predicts how genetic variants affect specific cell types and states rather than cells broadly, and DeepMind says it identifies nearly twice as many expression-changing variants as its prior best model.
Google DeepMind has released AlphaGenome 2, an upgraded AI model that predicts how non-coding DNA sequences affect gene activity across hundreds of specific human and mouse cell types. Using the model, researchers reproduced key findings about an inflammatory bowel disease gene variant, work that originally took immunologist James Lee's team at the Francis Crick Institute about 18 months of lab experiments, in roughly 30 minutes. The release follows DeepMind's earlier AlphaGenome Atlas, which mapped predictions for all 9 billion possible single-letter DNA changes but lacked cell-type specificity.