LRCP: identifying functionally related SNPs and genes from phenome-wide genetic-effect correlations using GWAS summary statistics

LRCP estimates phenome-wide genetic-effect correlation between SNPs in 452 UK Biobank GWAS, correcting LD and sample-overlap bias to recover 17 FDR 5% variant pairs.

• Why it matters: Naive Z-score correlations across many traits are confounded by linkage disequilibrium, sample overlap, and shared genetic correlation, causing 51–70% of null distal variant pairs to be falsely rejected.
• What they did: LRCP regresses Z-score cross-products on LD cross components using per-SNP heritability enrichment from LRCQ and LDSC-derived plug-ins, with exact standard errors for cross-GWAS dependence.
• The result: In simulations LRCP is unbiased (absolute bias ≤0.03) with 93–98% interval coverage and 5.0–5.7% type-I error; in Pan-UK Biobank it yields 113 strongest links (P < 10-5) and Reactome pathway sharing 4.9× random.

AI summary of the abstract · may contain errors · not medical advice · read the paper.