3.1 Analysing alteration in PSN upon transient association
Preliminary analysis of the constructed PSNs was performed to study the alteration of basic network parameters like average degree and strength, change in hub status, loss or gain of edges and centrality information. Supplementary Figure 1 shows the degree of all nodes in the dataset. The maximum degree of any node in all PSNs analysed in this dataset is 19. Nodes in the PSNs that have a very high degree are special hotspots with high connectivity that can alter the overall network when perturbed. These nodes in the PSNs are identified as hubs. From the degree distribution of the dataset, it is observed that nodes with degree equal to 11 crosses the 90th percentile. Henceforth, for this analysis, any node with degree greater than 10 is considered as a hub node.
The change of degree and strength at all sites between the PSNs of the bound and unbound forms of the proteins were next analysed. The alteration at the interfacial sites, primary contacts and non-interfacial sites were considered as the changes at, around and away from the interface, respectively. Supplementary Figure 2 summarises the changes observed across all sites in the dataset. The variation upon transient association are recognised by treating the unbound form as the reference structure and measuring all changes in the network when complexed. Hence, any edges that are unique to the unbound form are considered to be lost and those unique to the bound form are found to be gained. Likewise, altering the hub status of nodes in the networks are also noted. If the size of the protein is large and has a compact structure, there is an increased chance of finding more hubs. Loss or gain of numerous hubs can point to change in local conformation of sidechains. The information on gain and loss of edges and hubs in all cases is reported in Supplementary Table 1.
Across the dataset, a net gain in hubs is observed more: In terms of hubs, the net gain, loss and indifference are 437, 333 and 125 cases, respectively. Human serum albumin in complex with beta-2-microglobulin and IgG receptor (PDB ID 4K71) has the highest net gain in the number of hubs (23) across the dataset. However, taking percentage gain proportional to the length of the protein, the highest gain in hubs is observed in another IgG Fab fragment (PDB ID 1NSN) where there is a gain of 15 hubs in the 187-residue structure, which is more than 8% of the protein. Such a tremendous gain in hubs upon binding would suggest that the residues move closer to each other, and the structure would be more compact upon binding. On the other hand, the highest loss (33 hubs and 103 edges) is observed in the case of Glutathione S-transferase when bound to Jasmonic amino acid synthetase (PDB ID 5GZZ). Interestingly, 164 edges are lost whereas only 61 edges are gained and the loss of hubs from unbound form is not accompanied by any gain of hubs when bound, which suggest that there is a severe loss of connectivity in the protein as a result of protein binding. The information necessary to analyse each case in the dataset is provided in Supplementary Table 1.