For all functional enrichment analyses, 100 matched sets of variants (hereon ‘control variants’) were used as the null. These sets were generated with SNPsnap 76 using unique lead eQTL variants from different datasets as the input. Variants were matched for minor allele frequency, number of SNPs in LD (‘LD buddies’; r2 > 0.5), distance to the nearest gene, and gene density, allowing for maximum deviation of +/- 50% for each criterion. HLA SNPs (defined as falling between positions 25,000,000 and 35,000,000 on chromosome 6) were excluded from the analysis and all matched sets were non-overlapping with the input variants. 1000 Genomes Phase 3 European population was used as the genotype reference panel. Both target and control sets of variants were further expanded with their high-LD proxies (r2 > 0.8) derived from the European UK10k reference panel. These expanded sets form the basis for all subsequent enrichment analyses.
