Figure 3

One-shot WGD network duplication model and connected structure of the transcription-factor WGD network. Top panel: in order to test for significant large-scale structures in the organization of duplication subgraphs in the network, we employed a one-shot network WGD (null) model, assigning to each ancestral edge a random duplication subgraph with probability equal to its empirical frequency. The model can be described as a random coloring of the ancestral (collapsed) network links. Middle and bottom panel: significance of the connectedness of the transcription-factor WGD network scored by the one-shot model. For each one of 4000 model realizations, we considered the histogram of the number of connected components (middle plot) and of the size of the largest connected component (bottom plot) of the network involving transcription-factor DSs with two or more links), and compared it to the empirical case. Networks containing a single large connected component as in the empirical one are very infrequent in the random model (P-value 0.001 for number, 5E-4 for size of connected components).