we relax the constraint of requiring a gene to be expressed in both tissues, then the proportion of shared ASE effects is substantially lower (0.85 to 39%, mean 11%; fig. S22B). This finding represents the total probability of detecting a regulatory effect in another tissue, and highlights a high degree of apparent tissue specificity that derives from the fact that a gene expressed in one tissue is often simply not expressed in another. This is particularly pronounced in brain, where the large proportion of genes showing tissue-specific isoform expression in that organ (17) drives a lower degree of overall sharing. Whole blood and skeletal muscle are partial outliers, relative to other tissues, with lower sharing of eQTLs and lower average replication in other tissues (Fig. 2A, fig. S12, and figs. S20 to S22).