Network visualization and clustering of the pig transcriptome. A. Three-dimensional visualization of a Pearson correlation graph of data derived from analysis of pig tissues and cells. Each node (sphere) in the graph represents an individual probeset on the array and the edges (lines) correspond to correlations between individual measurements above the defined threshold. The graph is comprised of 20,355 nodes (probesets) and 1,251,575 edges (correlations ≥0.8). The complex topology of the graph is a result of groups of co-expressed genes forming cliques of high connectivity within the graph. Clustering of the graph using the MCL algorithm was used to assign genes to groups based on coexpression. By inspection of the underlying profiles, areas of the graph can be associated with genes expressed by specific tissue or cell populations. Plots of the average expression profile of genes in selected clusters are given on the right: B. profile of cluster 4 genes whose expression is restricted to brain and spinal cord; C. profile of cluster 7 genes whose expression is highest in blood; D. profile of cluster 10 genes whose expression is restricted to skeletal muscle; E. profile of cluster 22 genes whose expression is highest in the adrenal gland. MCL, Markov cluster algorithm.
Freeman et al. BMC Biology 2012 10:90 doi:10.1186/1741-7007-10-90