To provide a view of coordinated gene regulation arising from both cis and trans genetic effects and nongenetic sources, we inferred gene-gene coexpression networks within tissues. Studying each of the nine tissues in isolation and examining variation of gene expression across individuals, we linked pairs of genes that show correlated expression (top 1% of pairs by Pearson correlation) (14), revealing similar patterns of coexpression across tissue pairs (Fig. 6A). The median π1 statistic (fraction of true positive results) (26), estimating the total fraction of coexpressed gene pairs identified in one tissue that are replicated in a second tissue, is 0.44, ranging from 0.30 to 0.58 (Fig. 6A). Furthermore, the specific tissue pairs with higher overlap of coexpressed genes also have a greater overlap of cis-eQTLs than other tissue pairs, indicating a similar pattern of tissue relatedness underlying both results (P < 0.05 for correlation between similarity matrices of Figs. 2B and 6A). Thus, although coexpression networks primarily capture non-cis mechanisms, including trans regulation and environmental factors, the overall level of sharing and the specific patterns of tissue relatedness agree with the patterns observed from cis-eQTLs, which suggests that regulatory mechanisms beyond cis effects may be shared across tissues.
