Researchers from Harvard Medical School and the Center for Genomic Regulation unveiled popEVE, a proteome‑wide deep generative model that combines evolutionary data with human population variation to estimate missense variant deleteriousness across the proteome. The tool aims to accelerate rare‑disease diagnosis by prioritizing the most damaging variants when parental or population data are limited. Validation across thousands of families showed popEVE improves variant ranking in clinical settings and can function with single‑patient genomes—an advantage for resource‑limited clinics. The team published methodology and early clinical case examples, positioning popEVE as a practical triage tool for diagnostic labs and genomic medicine programs.