Analysis of eQTL effect sizes around the TSS demonstrated that, as a group, upstream variants had the strongest effects, while those within transcripts had the weakest effects (Extended Data Fig. 11b). This suggests that eVariants that are likely to affect transcription have stronger effects on gene expression levels than variants that are likely to affect post-transcriptional regulation of mRNA levels. A notable exception is splice site and stop-gained variants, which make up a small number of total eQTLs but have large effects on expression levels (presumably owing to nonsense-mediated decay). When genes are stratified by the number of tissues in which they are expressed, the average effect size decreases as the number of tissues increases, indicating that genes expressed in greater numbers of tissues are less likely to have eQTLs with large regulatory effects (Spearman’s ρ =−0.29, P ≤ 2.2×10−16; Fig. 3f).
