Vzyme: A Template-Based Method to Predict Reactions between Two Chemical Compounds

Abstract

Metabolomics plays an important role for linking genomes and cellular functions, since metabolites are components of cellular regulatory processes and their levels can be related to responses of biological systems to genetic or enviromental changes. Compared to genome, proteome, and transcriptome, however, metabolome is just beginning to be unraveled because of the difficulties of experimental measurements. Especially, secondary metabolites are chemically instable and have complicated structures and their pathways are less well understood. We have developed a virtual enzyme system (Vzyme), which constructs metabolite networks by predicting whether each chemical compound has a reactive connection with each other. Vzyme is a template-based search program utilizing chemical knowledge and consisting of two steps. In the first step, empirical knowledge of reactant-product relationships is acquired from the REACTION section of the LIGAND database and organized in the form of a template library. In the second step, a query compound pair is checked with the template library and if any template is found an edge is made between the pair. We applied this method to infer reaction pathways of phytosterols which are plant secondary metabolites.