pair (SNP-based model) is active (Fig. 2C). Both approaches show a U-shaped pattern, with high tissue specificity (activity in a single tissue) or tissue ubiquity (activity in all nine tissues) more common than profiles involving only a few tissues, despite many more possible combinatorial patterns for intermediate specificity. Notably, both methods indicate that more than 50% of all detected eQTLs are common to all nine tissues. Reassuringly, both Bayesian methods produce pairwise tissue sharing probabilities that show agreement with the non–model-based analysis (fig. S13). Figure S14 illustrates the value of the multitissue analysis for an example in which the tissue specificity of an eQTL supports NDRG4 as a candidate to influence QT interval in the heart (30). Despite superficial similarity between patterns of tissue-tissue gene expression and eQTL sharing, extensive comparisons of eQTL evidence to expression levels indicate that the tissue-specific patterns are only modestly correlated with average tissue expression levels (fig. S15).
