The large number of cis-eQTL signals identified in this study (90,885 unique sentinel marker–expression ID–tissue combinations), across ten brain regions, allowed us to explore the organization of cis-eQTLs in the human brain. We note that a gene that is ubiquitously expressed in human brain may still display region-specific patterns if, as recent studies illustrate26, cis-eQTLs act in a context-specific manner. We used unsupervised hierarchical clustering (Pearson’s linear dissimilarity measure applied to absolute z scores of all cis-eQTL signals across all ten brain regions) to group cis-eQTL signals and brain regions into clusters of interest (Fig. 1a). Consistent with expectation, this approach identified shared cis-regulation among cortical brain regions, and in particular between frontal and temporal cortex. This approach also identified a striking number of distinctively region-specific cis-eQTL clusters. Notably, this clustering pattern was not apparent from the results of unsupervised hierarchical clustering of equivalent gene expression data (all expression IDs targeted by cis-eQTL signals; Fig. 1b). This suggests a more nuanced relationship between gene regulation and expression, which allows a gene to be the target of different cis-eQTLs in different
