There was some redundancy among the set of SNPs examined through the GWAS catalog because different published reports can highlight different (but correlated) SNPs for the same phenotypes, and also because several phenotypes may show associations to some of the same sets of SNPs (e.g. autoimmune disorders with associations to HLA region SNPs and PTPN22). To insure that our conclusions were robust to both linkage disequilibrium (LD) among the cataloged SNP set and which of any correlated set of SNPs were retained for analyses, we conducted simulations in which we retained only one of any SNP set with r2>0.3 and for each replicate conducted 100 simulations conditional on MAF as for Figure 1 (see Materials and Methods for simulation details). The eQTL enrichment was preserved in LD-pruned sets of cataloged variants, for eQTL definition thresholds with p<10−4 or smaller.
