We used these modules to identify instances of coordinated variation in multitissue expression patterns across individuals. For each gene, we estimated the proximity of each individual’s expression pattern to the median expression pattern of each module, corresponding to a “module membership score” (Fig. 6F and fig. S31A). We then calculated a module membership score for each individual (Fig. 6G) (14). Cases where depletion in membership of one module was accompanied by a corresponding increase in membership of a different module are called “module switching events.” The majority of genes showed conserved multitissue gene expression patterns among individuals, remaining in the same module or switching between modules with similar expression patterns (correlation distance < 0.5). However, we identified 3965 genes (21%) that show switching between dissimilar modules (correlation distance > 0.5), which may have important biological consequences (fig. S31B). Using module membership scores as a quantitative trait, we searched for neighboring SNPs that are correlated with these module membership scores (Fig. 6H), which we call module-switching QTLs (modQTLs). Searching a window of 1Mbp around each gene in cis, we found a
