Eigenvector centrality based algorithm for finding a maximal common connected vertex induced molecular substructure of two chemical graphs

• Published in: Journal of molecular structure Vol. 1244; p. 130980
• Main Authors: Parisutham, Nirmala, Rethnasamy, Nadarajan
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 15.11.2021
• Subjects: Graph algorithm; Maximal common subgraph; Molecular similarity search; Substructure mining; Maximal common subgraph; Substructure mining; Molecular similarity search; Graph algorithm
• ISSN: 0022-2860; 1872-8014
• DOI: 10.1016/j.molstruc.2021.130980

•It uses of eigenvector centrality.•ONE DFS SEARCH is sufficient.•It gives large size maximal common connected subgraph. The physical and biological properties of a chemical molecule entity are related to its structure. One of the basic widely accepted principles in chemistry is that compounds with similar structures frequently share similar physicochemical properties and biological activities. The process of finding structural similarities between chemical structures of molecules helps to identify the common behavior of these molecules. A familiar approach to capture the structural similarity between two chemical compounds is to detect a maximal Common Connected vertex induced Subgraph (CCS) in their molecular chemical graphs. The proposed algorithm detects a maximal CCS by checking the induced property of the vertices which are collected by performing a DFS search on the tensor product graph of two input molecular chemical graphs. The DFS search will start from the node which has the highest eigenvector centrality in the tensor product graph. The significance of the proposed work is that it uses eigenvector centrality to predict the root node of the DFS search tree, so that the resulting sugraph gets more number of nodes (i.e. large size maximal CCS). The experimental results on synthetic and real chemical database, further ensure the competence of the proposed algorithm when compared with the existing works.