eQTN enrichment within exons is strongly supported by all three datasets while there is relatively weak evidence that the last exon is special.

The table displays the odds ratio estimates together with their corresponding 95% confidence intervals, as estimated by the empirical Bayesian model (see Methods) for each expression dataset (Illumina, Affymetrix, RNA-seq). Model 1 (TSS, intragenic) estimates the odds ratio that a SNP inside the transcribed region is an eQTN compared to a SNP outside the transcribed region (controlling for distance from TSS). For Model 2 (TSS, intron, exon) and Model 3 (TSS, intron, exon, last exon) we used the intron annotation as the reference: the reported exon and last exon odds ratio can then be interpreted as the relative odds that a SNP within these regions is an eQTN with respect to an intron SNP at the same distance from the TSS.