Matching the strongest cis-eGenes was performed as follows: for each of the 23 trans-eGenes g, each having Ng associated eVariants (10% FDR), we randomly selected a cis-eGene amongst the ten strongest cis-eGenes in that tissue, based on the P value of the strongest associated variant that also had at least Ng associated variants (10% FDR). We then selected the top Ng associated variants with this gene based on P value. Selecting 23 random cis-eGenes a single time yields unstable results, so we ran cis-eGene selection and enrichment 70 times with different selections. This was done for both random cis-eGenes and random selections amongst the strongest cis-eGenes. We rank-ordered the 70 trials for both promoters and enhancers based on average odds ratio enrichment relative to background. We then used the trial that was closest to median rank for plotting both promoter and enhancer enrichment results.
