Google DeepMind releases AlphaGenome 2 for single-cell DNA analysis
Google DeepMind has launched AlphaGenome 2, a model that predicts genetic variant effects in specific human cell types with improved accuracy.
Google DeepMind announced AlphaGenome 2, a next-generation DNA sequence model designed to predict how genetic variants affect hundreds of human cell types. The system uses a conditional model architecture that takes a specific biological state as an extra input alongside the DNA sequence. It analyzes individual cell states to compute gene activity, physical DNA packaging, and chemical switches on the DNA. The model was trained on more than 107 million individual cell profiles from 68 tissues in humans and mice.
In testing on the TenK10K dataset, AlphaGenome 2 identified nearly twice as many causal variants compared to the original AlphaGenome fine-tuned on single-cell data. It identified about 30 percent of causal variants and correctly predicted directional gene effects in 43 percent of cases. The model can also infer missing data from a single measurement, such as predicting chromatin accessibility based on gene activity. It computationally reproduced a finding showing that a variant drives IBD by overactivating the ETS2 gene in gut immune cells.
AlphaGenome 2 is available today for non-commercial research via the AlphaGenome API and is in private preview for commercial research. The model can generalize to entirely new single-cell experiments without needing to be retrained. It is not approved for clinical diagnostics.
Sources
- Google DeepMind BlogAlphaGenome 2: Deciphering how cells read DNA to better understand disease ↗
- The DecoderGoogle Deepmind shifts genomics AI from bulk tissue predictions to single cell type resolution with AlphaGenome 2 ↗
Every sentence of this story was checked against facts quoted from these articles. How we make the news
Spotted an error? Tell us
