Fig. 2: Comparing modular structure of structural networks.

Modules are cohesive subnetworks – nodes that make more connections to other members of the same module than to other modules. Here, we compare the modular structure of network with original weights and the same network with weighted, directed, and signed edges. Here, we examine modules estimated with a fixed resolution parameter (γ = 1) but explore the multiscale modular structure in the supplement. Co-assignment probability matrices for the inferred edge weight (a) and the fiber density matrices (b). c, d Consensus communities for both versions of weights. e Element-wise difference in module co-assignment. f System co-assignment matrix for reference. Black entries refer to pairs of brain regions that are assigned to the same system in both coarse and fine-scale system divisions. Gray entries are co-assigned to the same system for the coarse division, only. Comparison of laterality (g) and modularity (h) of detected modules (N = 1000 for each box plot). i Alignment of modules with respect to coarse and fine-scale system partitions. Each point in panels g–i represents a partition from one of 1000 runs of the Louvain algorithm for optimizing modularity. Box plots, shown in red and overlaid on data points in e and f, depict the interquartile range (IQR) and the median value of the distribution. Whiskers extend to the nearest points ± 1.5 × IQR above and below the 25th and 75th percentiles. Asterisks indicate a significant difference between box plots (N = 1000 for each box plot).