Maps GWAS hits to candidate genes using Open Targets locus-to-gene scoring instead of the unreliable nearest-gene heuristic. Pulls associations from GWAS Catalog and Open Targets Genetics, filters by genome-wide significance (p < 5e-8), and ranks genes by L2G scores that integrate eQTL, chromatin interaction, and distance evidence. L2G > 0.5 is considered strong prediction. Useful when you need to go from "what SNPs are associated with type 2 diabetes" to "what genes are actually implicated" with some statistical rigor. The skill emphasizes looking up real data rather than guessing, which is the right approach since GWAS catalogs update constantly and LD patterns differ across ancestries.
npx skills add https://github.com/mims-harvard/tooluniverse --skill tooluniverse-gwas-trait-to-gene