Optimal paths in residue networks identify communication pathways in proteins

Official URL: http://192.168.1.20/record=b1293852 (Table of Contents)

Navigation of information flows in networks is studied. As real-life systems, residue networks constructed from the coordinates deposited in the protein data bank are targeted. The cost of the navigation between neighbors are measured by residue-residue interaction potentials. By constructing all paths between initial/target nodes according to selected criteria, structurally and/or functionally important residues in the network are implicated. In particular, strong paths that minimize the weights along all possible pathways are found to differentiate between the functional nodes in protein families with high overall structural similarity, but low sequence similarity scores. To determine factors that drive the usage of strong paths in the network, a biased random walk scheme is deviced where the probability of edge selection is based on a balance between the knowledge of the location of the destination and the energy of interaction with the immediate neighbors. Since long range communication between two distantly placed functional regions in the protein calls for the gradient of information flow, strong paths emerge by satisfying the competition of local and global knowledge while navigating along the structure.