It should be noted that the organization of the human brain transcriptome described here could not have been revealed by standard methods, such as analysis of differential expression. This analytical approach seeks to identify genes that are expressed, on average, significantly higher or lower in one group versus another, with greater significance being attributed to genes for which the ratio of inter-group to intra-group expression variance is maximized. In our study, each dataset was comprised of a single group consisting of control samples from a specific brain region. In each group, the relative representation of specific cell types or functional processes in each sample was unknown. Therefore, there was no basis for identifying genes that were differentially expressed in these data sets, as there were no groups to compare. Furthermore, because the end result of differential expression analysis is a list of genes, each of which has been deemed significant in isolation, a systems-level understanding of such findings relies exclusively on post hoc analyses to connect individual genes in a broader functional framework. In contrast, analysis of gene coexpression relationships reveals the inherent organization of the transcriptome as the appropriate biological framework in which to consider additional post hoc analyses.
