Google DeepMind has introduced AlphaGenome 2, an AI model designed to predict how DNA changes affect specific types of cells and help researchers investigate the biological mechanisms behind disease. Alphabet ($GOOGL) CEO Sundar Pichai (@sundarpichai) highlighted the distinction between a shared genetic blueprint and the different ways cells use it. A heart muscle cell and an immune cell can respond differently to the same DNA sequence. DeepMind trained the model using more than 107 million individual cell profiles from human and mouse samples. Unlike its predecessor’s broader tissue predictions, the new system can target a particular tissue, cell type or biological state. In the TenK10K benchmark covering 28 primary immune cell types, AlphaGenome 2 identified 30% of causal variants, compared with 16% for the earlier model fine-tuned on single-cell data. Its reported performance on predicting the direction of gene activity also improved, from 23% to 43%. DeepMind demonstrated the system by reproducing an existing finding linking a DNA variant to increased ETS2 gene activity in intestinal immune cells associated with inflammatory bowel disease. That was a computational result, not a new treatment or a clinical trial. The model is available for noncommercial research, with commercial access in private preview. It has not been approved or validated for diagnosis. Its practical value will depend on helping scientists identify stronger leads for laboratory testing.