Fig. 1

The complexity of the A. thaliana regulatory network, two clustering strategies and the work flow of CoReg. a The complexity of regulatory A. thaliana network. Each node represents one gene in the network and each edge represents an interaction between one TF and its target. We classified the nodes into three categories based on the degree: 1) triangle, in-degree = 0; 2) rectangle, in-degree >0 and out-degree >0; 3) circle, out-degree = 0. c CoReg uses a clustering strategy different from existing clustering method. Typically, the network modules that normal clustering algorithm identifies are shown on the left. However, if there are two genes which share many targets and regulators in common, they are most likely to be the actual co-regulators (shown on the right, gene A and gene B) CoReg is designed to work on the clustering problem on the right. b The brief work flow of CoReg starting from input (a regulatory network). Red nodes in the second step represent common target (for out-similarity) or regulator (for in-similarity) for the pair of nodes in the middle. CoReg adds up the incoming similarity and outgoing similarity and then calculates a distance matrix. Next, distance matrix is used as the input to hierarchical clustering. In the last step, dynamic tree cut is performed to obtain final module assignment for each node. d flowchart of CoReg analysis