Enzyme simulations are notoriously challenging as their large size prevents modeling the whole enzyme with highly accurate but computationally expensive quantum mechanical (QM) methods. Currently, only the chemically important “active site” residues/substrates/cofactors are modeled with QM-methods. Determining which species should be included in the QM-model remains ambiguous, and current practices for model construction typically suffer from human bias and unclear methodology. Our lab has been developing the RINRUS (Residue Interaction Network ResidUe Selector) toolkit which seeks to provide a solution by automating the design of QM-models using quantifiable residue interaction networks. To illustrate the utility of RINRUS, a series of QM-cluster models were constructed of the enzyme catechol-O-methyltransferase (COMT). The reactant, product, and transition state of the methyl transfer reaction catalyzed by COMT was computed for a total of 538 models, and the resulting free energies of activation and reaction were used to evaluate model convergence. RINRUS-designed models with only 200 – 300 atoms are shown to converge. k-means clustering was used to group models that predict similar reaction thermodynamics, allowing the identification of characteristics that yield efficient, accurate models. The results from this case study will be applied to other enzyme systems so that RINRUS may serve as a cornerstone for improved and automated cheminformatics-based enzyme model design.

Incorporating cheminformatics into the design of quantum mechanical cluster models

ACS Spring Meeting 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.