Fig. 1

Assessing the evolutionary divergence of human and mouse brain networks. a Generation of human and mouse brain networks. RNA-seq expression data derived from 12 brain regions from the GTEx dataset and 7 brain regions from publicly available mouse data were used to create gene networks in each brain region using rWGCNA. These networks were hierarchically merged to generate a consensus whole-brain network. Co-expression modules were generated for each network and assessed for evolutionary divergence. Each module was assessed for transcriptome (also referred as “co-expression” or “module”) divergence as well as sequence divergence. b For each module in each brain region of the human and mouse brain reference networks, we assessed module reproducibility (preservation) in multiple independent test datasets derived from the same region in human, non-human primate (NHP), and mouse. For each test, a composite Zsummary (Zsum) statistic representing preservation was generated. For both human, NHP, and mouse, the upper quartile (UQ) of all test Zsum scores was calculated to generate a final preservation score in their respective species (hZsum, pZsum, mZsum). Modules with a same-species Zsum UQ > 5 are considered to be robust and reproducible. To assess divergence between human and mouse, a module divergence score was calculated as follows: (hZsum/mZsum) − 1. This framework was also applied using list of genes generated through alternate methods to assess how their co-expression structure has changed across evolution. c We subsequently used this framework generating evolutionary divergence scores to assess the transcriptomic divergence of brain regions and cell types and determine the genes which underlie this divergence. d Using these data, we create a resource to guide disease modeling. For example, we assess which human disease genes’ co-expression is not preserved in mouse and assess to what extent different model systems capture human co-expression patterns