Figure 2

Degree distribution, weight distribution and topological analysis. A) Degree distribution of nodes (mRNAs/miRNAs); blue refers to the complete set and red to the fraction of nodes included in the consensus modules. The variance of modular nodes was significantly different (Bartlett test, P-value = 1.4E-39) and with higher mean value, which is translated into more hub nodes favored during module construction. B) Calculation of the characteristic path length (CPL) of the complete network, of the network after removing the 186 age-dependent nodes, of the network after removing the 429 consensus modular nodes, of the network after removing the top 429 hub nodes and of the network after removing randomly 429 nodes (mean value after 100 runs). If the nodes removed are important mediators for network communication CPL will increase. C) Weight distribution of the complete multi-layer network with bars representing the mean edge weight value as calculated from all combinations of mRNA/miRNA expression experiments and error bars depicting the confidence intervals. Each bar represents a value range named after the upper limit. All weight values were adjusted to ‘1-weight value’ due to the fact that DMSP algorithm constructs modules by promoting edges with weights closer to zero. As shown, with cutoff value ≤ 0.4, only 8% of the complete interactome (~5,300 edges) changed substantially during lifespan D) Weight distribution of the edges included in the final consensus modular topology. With cutoff value ≤ 0.4, 1,270 edges were included in the consensus modular topology which represent the 24% of the total age-related relations (one-sided Fisher exact test, P-value = 2.2E-16).