such an approach offers several advantages over the conventional methods used to study neurological or neuropsychiatric disorders. First, differential network analysis can suggest dysregulation at the level of modules, which may implicate specific cell types or functional pathways. Although pre-existing classification systems such as gene ontology49 are essential tools for microarray studies, these systems currently do not provide adequate context in regards to tissue and cellular specificity of gene expression patterns. In comparison, modules of coexpressed genes that have been identified in an unsupervised fashion in a biological system of interest possess immediate functional relevance. Second, in contrast to studies of differential expression, differential network analysis effectively normalizes gene expression levels relative to their primary source(s) of variance. This practice can help control for sample differences resulting from biological factors (for example, differences in the absolute quantities of specific cell types) and technical factors (for example, differences between dissections or sample quality) that can influence the results of differential expression studies. Third, in comparison with genomic studies, analysis of the transcriptome in affected brain regions is more proximal to function than analysis of DNA sequence.
