Model-based clustering of bimodal gene expression identifies cohesive clusters in 19 tissue types. Heat map representation of posterior pairwise probabilities for classification of tissue phenotype. Left column: classification with 1265 bimodal genes. Right column: classification with 300 bimodal genes translated into extracellular matrix or plasma membrane proteins. Top row: Model-based clustering, identifies all tissues distinctly. Middle and bottom rows: Kmeans and hierarchical clustering classify samples into three/four tissue types: brain, cardiac and skeletal muscle and remaining tissues. Blue, green, yellow, orange and red regions of color bar indicate ovary, stomach, small intestine, pancreas and thymus tissue samples respectively. Tissues in the heat map were ordered according to decreasing sample size from left to right.