Workflow overview.

<p>In each dataset, the following workflow was applied. 1. The dataset was used as a starting point to construct a gene coexpression network based on Topological Overlap between genes. TO determines similarity between gene expression profiles taking into account a systems level context. The network was next hierarchically clustered, resulting in a cluster dendrogram. 2. Using DynamicTreeCut algorithm, branches were identified in the dendrogram, leading to identification of gene coexpression modules. 3. Genes in each module were further prioritized by intramodular connectivity, providing a distinction between lowly and highly connected genes. The entire workflow was repeated independently for 82 datasets, resulting in a set of gene coexpression modules in each of them.