Graph-Based Bioinformatics Mining Research and Application

Abstract

In the biological technology domain, the biologist discovered frequent sub graph mining may reduce the price of structure match experiment in the protein gene structure match experiment. Molecular model can be abstracted as a set of graph, it usually needing to find out the particular molecular structure in biomedical testing. Therefore, research on the frequent sub graph mining has the important significance of theory and the application value. Our contribution in this paper includes: (1) based on the usual frequent sub graph definition, we propose a novel definition of frequent sub graph mining, (2) propose a nove graph canonical form to determining graph isomorphism and avoid the NP-Complete problem of sub graph isomorphism, (3) Can efficient enumerated candidate subgraph's supporty and frequency by maintaining an embedding set. At last, performance study indicates that FSubgraphM can effectively discovery the frequent induced sub graph from the database carcinogen, it also can form some interesting association rules from the frequent sub graphs which has some theoretical and practical significance for Bioinformation Data Mining and Its Application.