We compared networks using various preservation statistics that can be grouped in two main categories: Density-based preservation statistics are used to determine whether the genes in a reference module remain highly connected in the test network.Connectivity-based preservation statistics assess whether the overall connectivity pattern between genes in a reference module is similar in the reference and test networks. Network statistics used to assess preservation of density and connectivity are described in the supplementary text. Within each category (density or connectivity), composite module preservation statistics are constructed to summarize changes in module preservation. In detail, the comparison of network preservation in the reference and test networks is based on a permutation-based approach. The permutation approach implemented in the WGCNA package (module label permutation in the test network) shows a strong dependency on module size in our cohort (fitting index r2 > 0.95 based on a quadratic model). Thus, as an alternative, we performed 1,000 permutations of disease status for the final cohort analyzed for co-expression (278 control and 254 schizophrenia samples), followed by generation of gene co-expression networks and estimation of
