Systems Level Analysis of Systemic Sclerosis Shows a Network of Immune and Profibrotic Pathways Connected with Genetic Polymorphisms
Figure 1
Schematic of the analysis pipeline for integrative analysis of multiple SSc skin datasets.
(A) Each microarray dataset (Milano et al., Pendergrass et al., and Hinchcliff et al.) was independently clustered by WGCNA into gene coexpression modules (colored circles). Each module is a set of genes that was highly correlated within a dataset. (B) Modules were compared across datasets using a novel procedure (MICC) to determine which were approximately conserved across all three datasets. The network in (B) is called the information graph and encodes the nontrivial overlaps of modules across datasets. Triangles in this network correspond to approximately conserved modules across all three datasets. Communities in this network (dotted ovals) represent collections of modules that are conserved together and thus have similar biological function. Note that communities in the network can overlap (e.g. module P1 in the schematic belongs to two communities). (C) Genes derived from the module communities are called consensus genes and were used for downstream bioinformatics analyses including gene ontology enrichment analysis using the g:Profiler tool, testing for intrinsic subset-specificity, and functional interaction network analysis using the IMP functional network. Each of these downstream analyses is independent and complementary.