TransMiner: Biotransformation Prediction

Minesh Upadhyaya1, Imran Shah2, Daniel McShan, Weiming Zhang, UCHSC;, UCHSC

TransMiner: Biotransformation Prediction Upadhyaya, M. McShan, D. Shah, I. MOTIVATION One of the main challenges to understanding the pathways in a living system is identifying new metabolic capabilities. We are developing TransMiner, a symbolic computational approach for inferring novel biochemical functions. METHODS Data on thousands of enzyme-catalyzed transformations and associated metabolites is available in BioCyc and KEGG. This information is invaluable for learning about the fundamental nature of metabolism. We abstract biotransformations expressively in terms of mappings between the chemical graphs of metabolites. We have developed a novel sub-graph isomorphism-based algorithm to search the detailed representations of known biotransformations to induce biocatalytic "rules". These biocatalytic rules represent the symbolic functions of enzymes. We can also apply these rules to infer novel possible biotransformations that can be catalyzed by an enzyme. Our biochemical inference algorithm and chemical knowledge are developed in Lisp, which is a flexible symbolic computation environment. RESULTS This poster presents and overview of our approach and summarizes our results for a large number of biocatalytic functions.