Figure 1

Outline of the WGCNA analysis. Phase 1) Obtaining pure microglia datasets. Transcriptome datasets were obtained from microglia of aged, accelerated aged, App-Ps1 transgenic (Alzheimer’s Disease model), Sod1 transgenic (Amyotrophic Lateral Sclerosis model), and i.p. LPS injected mice (acute activation). Each dataset contained its own control. Phase 2) Co-expression network formation. Co-expression networks were generated for 7512 genes of the indicated transcriptome datasets. Average linkage hierarchical clustering was applied to the topological overlap matrix and branches of highly correlating genes were formed, which were cut and assigned a color. Primed microglia networks were combined into a consensus network that represents the commonalities in the gene expression profiles of the individual primed microglia networks. Phase 3) Differential ME expression. For each module the Module Eigengene (ME) was calculated, which represents the expression profile of the module. A Kruskall Wallis between group test was applied to determine if ME values were significantly different between conditions, to find modules that were related to phenotype. The consensus primed microglia blue modules and acute red module are depicted as a box-plot containing the distribution of the ME values across the samples of each particular condition. Phase 4) Overlap between modules. The Fisher’s exact test was used to determine the significance of the overlap between modules from different model systems. Phase 5) Annotation of the modules. Modules were annotated using WebGestalt for GO and KEGG analysis. Phase 6) Comparison of core profiles. The correlation of each gene to the module EigenGene (kME) values was calculated for all genes in the analysis of the consensus blue priming module. These consensus primed microglia derived hub genes were subsequently compared to the acute activation network to find genes generally associated with activation, uniquely with primed microglia, or uniquely with acute LPS activation.