New Theory and Ontologies for Quantum Mechanical Cluster Modeling of Proteins and Enzymes

Atomic-scale computer modeling of enzymes is part of multi-domain and multibillion-dollar research efforts to study human health, toxicology and biodegradation, biocatalysis, and to advance our fundamental understanding of protein structure and function. Despite the widespread use and success of computational enzymology, quantitative relationships between the composition and size of the models and accuracy of simulations are still poorly understood, and comparison of methodologies is nearly impossible. This project will address some of the unanswered fundamental questions about how to design accurate atomic-level enzyme models, even more so about how to use published data from disparate enzyme families, different software packages, quantum chemistry methodologies, and experimental studies. The team will design data management infrastructure and data cleaning requirements for computational enzymology studies. When completed, the data management system will eliminate many severe limitations to the ability to systematically accumulate big amounts of enzyme simulation data and subsequently formulate structure-activity relationships via expert systems and machine learning. The technical feasibility of RINRUS as a high-throughput computational enzymology platform will be tested via 1) bioremediation and bioconversion studies of lignin with cytochrome P450 and 2) a kinetic study of glycosylation in Cel7A glycoside hydrolase. This effort is in collaboration with groups at the National Renewable Energy Laboratory and serves the DOE mission of improving energy security and sustainability. An automated pipeline for computational enzymology work, from X-ray crystal structures to reliable computational models, will vastly decrease project turnaround time for an entire modeling community. With a successful Phase I, Phase II efforts will focus on hardening inhomogenous cavity potentials for enzyme models, taking on a substrate/protein engineering design role with NREL collaborators, and further automating and simplifying atomic-level biomodelling workflows integrated in the RINRUS and Q-Chem packages. Lower barriers to using atomic-level enzyme modeling, which increases the audience and sales for Q-Chem, Inc. Higher-throughput R&D for biofuels, biocatalysis, renewable energy industries. Indirect research benefits to biomodelers in the drug design, national defense, and toxicology communities.