Prediction of interface residue based on features of residue interaction network extracted using shapley value centrality measure
Babita Pandey, Sarneet K Chawla, Devendra Kumar Pandey · 2018
Protein-Protein interaction plays an important role in the life processes. Molecular mechanisms of the related processes can be better understood with the help of interface prediction. In this work, we use game theory concept of Shapley value to analyse the spatial relationship between residues in residue interaction network. Four features are extracted from network using shapley value and given as input to ACO for optimization. Our experiment shows that optimized feature set, significantly improves the result of normal classifier and accuracy from 80% to 85%. These findings are useful for identifying protein-like complex networks. The presented results suggest that the feature selection by Shapley value and optimization by ACO improves the classification of protein structure at great extent less computational complexity.