What problem does it solve?
Most GWAS-based risk tools rely exclusively on European reference populations, missing variants whose effect sizes differ substantially in other ancestries (e.g., KCNQ1 rs2237892 has near-null effect in Europeans but OR=1.31 for type 2 diabetes in East Asians). This Skill infers genetic super-population ancestry directly from a raw genotype file and compares ancestry-specific GWAS effect sizes against European reference estimates.
Core Features & Use Cases
- Ancestry inference: Hardy-Weinberg log-likelihood scoring across ~72 ancestry-informative SNPs assigns one of five 1000 Genomes super-populations (AFR, AMR, EAS, EUR, SAS) with a soft posterior probability and confidence level; it abstains when fewer than 30 panel markers match.
- Ancestry-stratified risk comparison: Computes combined odds ratios per disease using ancestry-specific effect sizes versus European reference ORs, plus an exploratory Ancestry Elevation Score (AES) highlighting where ancestry diverges from European predictions.
- Recessive compound models: Handles loci like APOL1 (kidney disease) and HFE (hemochromatosis) with biologically correct recessive models rather than per-allele additive scoring.
- Use Case: A user with a 23andMe file asks whether their South Asian genetic background changes their type 2 diabetes risk; the Skill infers SAS ancestry and reports T2D with an elevated AES alongside per-variant detail and cited PMIDs.
Quick Start
Ask the AI to run the ancestry-risk-profiler demo to generate an ancestry-stratified disease risk report from the built-in synthetic South Asian 23andMe profile.
