AnnotateMissense: a genome-wide annotation and benchmarking framework for missense pathogenicity prediction
Researchers have developed AnnotateMissense, a new framework for predicting the pathogenicity of missense genetic variants. This system integrates a wide array of data, including population frequency, evolutionary conservation, and features derived from protein language models like AlphaMissense and ESM. Benchmarking against over 130,000 ClinVar-labeled variants showed that an XGBoost model using 303 features achieved a high performance with an MCC of 0.9411. The framework was then applied to over 90 million variants to generate pathogenicity scores. AI
IMPACT Provides a new tool for interpreting genetic variants, potentially accelerating research in genomics and personalized medicine.