The overlap of the modules described above among all networks is depicted in Figure 5. As seen here, gene coexpression modules do not exist in isolation, but rather comprise a meta-network whose higher-order structure reflects relationships between modules. To explore the relationships among identified modules within each network, we calculated the Pearson correlation coefficients between all possible pairs of module eigengenes (the resulting ‘module eigengene networks’41 are presented in Supplementary Fig. 6 online). These results confirm and extend many of the module characterizations described above (Table 1 and Supplementary Note). For example, two modules with characteristics of PVALB-positive interneurons (M6A and M17A) were significantly correlated in CTX (r = 0.57, P = 5.8 × 10−7; Supplementary Fig. 6). In CN, a module of genes coexpressed in the SVZ (M13C; see below and Fig. 6) was significantly correlated with a module enriched for astrocyte markers (M15C, r = 0.60, P = 8.4 × 10−4; Supplementary Fig. 6). In both CTX and CB, modules related to synaptic function (M10 and M14) were significantly correlated (r = 0.59, P = 1.3 × 10−7
