The abstract is read at the publisher; the summary is JClub's.
AlphaGenome 2 (AG2) substantially improves cell-type-specific prediction of regulatory activity and variant effects from a 1-Mb DNA sequence, doubling recall of causal single-cell eQTLs at 90% precision.
- Why it matters: Non-coding variants often act through regulation in specific cell states, but single-cell atlases lack enough donors to link variants to expression directly, and existing models struggle to separate related states.
- What they did: AG2 is trained on organism-wide human and mouse single-cell atlases plus bulk functional genomics, conditions on a target tissue/cell state, imputes epigenomic tracks, and predicts multimodal variant effects without fine-tuning.
- The result: AG2 outperforms fine-tuned AlphaGenome 1 and specialized target-dataset models, achieves state-of-the-art enhancer-gene linking, and generates enhancer-gene links and scores for 7 million common variants across hundreds of human cell types.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.