Simple Pattern-only Heuristics Lead to Fast Subgraph Matching Strategies on Very Large Networks
A wide range of biomedical applications entails solving the subgraph isomorphism problem, i.e. finding all the possible subgraphs of a target graph that are structurally equivalent to an input pattern graph. Targets may be very large and complex structures compared to patterns. Methods that address this NP-complete problem use heuristics. Their performance in both time and quality depends on a few subtleties of those heuristics. This paper compares the performance of state-of-the-art algorithms for subgraph isomorphism on small, medium and very large graphs. Results show that heuristics based on pattern graphs alone prove to be the most efficient, an unexpected result.
KeywordsSubgraph isomorphism Networks biology Search strategy
- 1.Bonnici, V., Busato, F., Micale, G., Bombieri, N., Pulvirenti, A., Giugno, R.: Appagato: An approximate parallel and stochastic graph querying tool for biological networks. Bioinformatics 32(14), 2159–2166 (2016). Cited by: 2Google Scholar
- 4.Carletti, V., Foggia, P., Saggese, A., Vento, M.: Challenging the time complexity of exact subgraph isomorphism for huge and dense graphs with vf3. IEEE Trans. Pattern Anal. Mach. Intell. PP(99), 1 (2017)Google Scholar
- 5.Giugno, R., Bonnici, V., Bombieri, N., Pulvirenti, A., Ferro, A., Shasha, D.: Grapes: a software for parallel searching on biological graphs targeting multi-core architectures. PLoS ONE 8(10) (2013). Cited by: 10Google Scholar
- 9.Michael, R.G., David, S.J.: Computers and Intractability: A Guide to the Theory of NP-Completeness, pp. 90–91. WH Free. Co., San Francisco (1979)Google Scholar
- 13.Ullmann, J.R.: Bit-vector algorithms for binary constraint satisfaction and subgraph isomorphism. J. Exp. Algorithmics 15, 1.6:1.1-1.6:1.64 (2011)Google Scholar