Gene expression clustering reveals mouse protein interactome modules and fuzzy relationships among mouse cells and tissues. Heat maps showing clusters of mouse gene expression data (GSE10246) identified using (A) hierarchical clustering and (B) AutoSOME clustering. (C) Protein interactome  divided into subnetworks corresponding to co-expression clusters identified by AutoSOME. (D) Fuzzy cluster network of cell/tissue types in GSE10246. Nodes represent individual cell/tissue types (labeled with first word of each sample name only), node colors correspond to different clusters, and increasing edge thickness and opacity reflect increasing frequency of co-clustering between any given pair of nodes over all ensemble iterations (see ). (E) Expression data of four cell/tissue types from GSE10246 superimposed onto the ten largest subnetworks from panel C (Stomach = GSM258771; Lymph Node = GSM258691; Cerebral Cortex = GSM258635; Embryonic Stem Cell = GSM258658). All expression data are log2 scaled and median centered. In panel B, all clusters are ordered by decreasing cluster size, and the yellow-cyan color scale is identical to panel A. In panels A and B, all arrays (cell/tissue types) are horizontally ordered the same as the GSE10246 data set.
Morris et al. BMC Bioinformatics 2011 12:436 doi:10.1186/1471-2105-12-436