Once the PLI matrix has been converted to a graph, the next step is to characterize the graph in terms of its clustering coefficient C and characteristic path length L. The path length L was calculated as ‘harmonic mean’ distance between pairs as described by [27], making it possible to deal with vertices that are not connected. The values of C and L for every k were compared to 1000 random surrogate matrices, generated as described by [28], by calculating the ratio between the C and L of the patient and the surrogate data (referred to as C-s and L-s). To analyze the small world characteristics of the network we used the measure of small world index S [29], which is defined as S = (C/C-s)/(L/L-s). A network can be defined as a small world network if C/<C-s>≫ 1 and L/<L-s> ∼1, which means that any value of S greater than 1 is account for small world network.
