Using Boca and Leek’s FDR regression to leverage H3K27ac data with asthma GWAS p-values.
(A) (-log10) adjusted p-values from BL against raw (-log10) FDR values coloured by average value of q (H3K27ac fold change value). (B) Box plots of (-log10) p-values in the discovery GWAS data set and the UK Biobank data set for the 655 SNPs that were FDR significant in the original GWAS (“before”) and 5 newly significant SNPs after applying BL (“after”). Black dashed line at genome-wide significance threshold (5 × 10−8). (C) Sensitivity and (D) specificity proxies for the results. Sensitivity proxy is calculated as the proportion of SNPs that are FDR significant in the UK Biobank data set that are also FDR significant in the original GWAS or after applying BL. Specificity is calculated as the proportion of SNPs that are not FDR significant in the UK Biobank data set that are also not FDR significant in the original GWAS or after BL.