bioresearch-gwas-harmonizationlisted

# BioResearch Agent — GWAS Harmonization Skill ## Capability Harmonizes GWAS summary statistics across multiple ancestries for cross-ancestry MR: 1. **Cross-ancestry GWAS simulation** — generates per-ancestry GWAS summary statistics with realistic allele-frequency drift (AFR < EUR < EAS/SAS/AMR), effect-size heterogeneity, and deliberately injected data-quality issues (allele swaps, strand confusion, palindromic SNPs). 2. **Allele harmonization** — 5-case alignment: direct match, effect-allele swap, strand flip, palindromic SNP removal, and unmatched SNP exclusion. Standardizes all ancestries to a common effect-allele / non-effect-allele convention. 3. **Strand ambiguity resolution** — removes palindromic SNPs (A/T, G/C) near EAF = 0.5 where strand orientation cannot be inferred, using a MAF-distance threshold (default ±0.01). 4. **Allele-frequency divergence** — computes pairwise Fst-like statistics across ancestries to quantify genetic divergence and flag SNPs with extreme AF differences (a source of MR bias). 5. **Cross-ancestry signal overlap** — identifies genome-wide significant SNPs shared across ancestries vs ancestry-specific, informing instrument selection and portability assessment. Returns harmonized per-ancestry GWAS + AF comparison + overlap report, not a causal claim. ## Run ```bash bioresearch run gwas-harmonization --n-snps 500 --n-causal 30 --seed 42 --output-dir outputs/gwas-harmonization ``` ## Outputs (in `--output-dir`) - `gh