Figure 3

Status-specific ncRNA-mRNA network comparison based on parameters describing network topology. The size of the network (A), the average shortest-path length (B) and the average degree (C) have been calculated for each network. The node degree distribution of the three networks follows a negative power law (D), indicating a scale-free topology and the arguments are very similar in the three cases (−1.439 for relapse, −1.431 for remission and −1.484 for controls). The frequency distributions of the shortest path-length of the three networks have also been plotted (E). Finally, the similarity of the three networks has been estimated with a Principal Component Analysis (PCA). The first two components accounting, respectively, for 88.76% and 11.07% of the variability have been plotted (F). For the PCA, the values of the following network descriptor parameters have been used: number of connected components, network centralization, characteristic path-length, average degree, network heterogeneity and the argument of the power law fitting the node degree distribution.