2.4 Global network and structure comparison
Structures in their bound form are aligned with the corresponding structures in their unbound form and information of structural deviation (RMSD) between all pairs of chains are obtained. TM-align is used to compute the RMSD, which is a measure of Cα deviation between topologically equivalent residues. This is followed by global network comparison of the same equivalent residues using the network dissimilarity method discussed by Gadiyaram and coworkers.
The method of computing network dissimilarity involves storing the PSN as an adjacency matrix which is normalised to a Laplacian of the network graph followed by its spectral decomposition to obtain the eigen values and eigen vectors. A Forbenious norm of the difference between the adjacency matrices of the two structural networks provides the edge difference score (EDS). A difference in the local and global clustering of residues in the network is captured using the Edge-weighted cosine score (EWCS) and the correspondence scores respectively. The component scores are then used to compile the Network Dissimilarity Score (NDS) between the compared networks. Identical networks would have a NDS of zero. No two networks can have a dissimilarity greater than\(\sqrt{}3\).