EnzFIND: Method to identify enzymes for promiscuous biochemical reactions

Enzyme promiscuity enables some enzymes to catalyze multiple reactions. This ability of enzymes will contribute to the underground metabolism of an organism. Comprehensive listing of all reactants, catalyzed by an enzyme or identifying an enzyme set that can catalyze a reaction pose significant challenges. Such information if available, not only facilitates metabolic annotation and genome scale model reconstruction but also allows enzyme exploration for designing synthetic biochemical reactions. This article presents EnzFIND, a method to find all enzymes that can catalyze a reaction on similarity of molecular features such as molecular properties and signatures between the enzyme’s native reaction and probe reaction.EnzFIND is evaluated for its ability to score and identify reactions that may be catalyzed by the same enzyme using receiver operating characteristic curves. EnzFIND score with a threshold cutoff of 0.52 achieves a maximum accuracy of 0.95 in identifying promiscuous enzymes. The area under the curve value of 0.90 in true positive and false positive rate plot establishes EnzFIND as a strong classifier. Comparative assessment suggested that EnzFIND outperformed other stateof-the art tools. EnzFIND’s utility is demonstrated by analyzing reconstructed metabolic model of Escherichia coli. From a list of 140 reactions which are yet to be included within the Escherichia coli genomic scale metabolic model, the method successfully predicted 112 reactions. In addition transformations for 17 reactions are correctly reported. Only eleven reactions are erroneously assigned.EnzFIND enables identification of promiscuous reactions for a given enzyme based on both reaction similarity and substrate variability and can serve as a tool for defining underground metabolism.