Dr. Lance Westerhoff will be presenting a poster concerning the Company’s recent work in QM-based x-ray refinement at Chemical Computing Group’s 2013 North American User Group Meeting in Montreal, QC.
Title: Phenix/DivCon meets MOE: how Crystallographers and Computational Chemists are using MOE/DivCon in modern, quantum mechanics-based X-ray refinement
Authors: Oleg Borbulevych and Lance M. Westerhoff
Abstract: In conventional refinement, the geometry of the ligand within the active site is modeled according to the practitioner’s beliefs as expressed in the form of stereochemical restraints provided by the ligand library or CIF file. Further, metal centers, bound species, and so on can be difficult to capture correctly without significant human intervention. A quantum mechanics (QM) functional can be used in “real-time” over the course of the refinement as an alternative to stereochemical restraints and as a more accurate way to model protein/ligand geometry. Since the QM procedure does not utilize the CIF for refinement, it does not make a priori assumptions about the ligand structure and chemistry. Furthermore, it is able to address metal coordination spheres and other exotic or unusual protein/ligand chemistry situations. We have implemented an efficient method for routine application of semiempirical QM (SE-QM) to refinement by integrating our linear scaling, SE-QM engine DivCon with the PHENIX crystallographic package. The QM protocol provides an unprecedented opportunity to objectively reveal structural details of a ligand taking into account changes due to binding and these advantages have significant implications for high-throughput crystallography whether in an industrial or academic setting. Further, since the method does not require as much esoteric knowledge and experience as traditional methods, it could be used by a wider variety of scientists in the field.
We report refinement results for 50+ structures chosen from Protein Data Bank (PDB) and refined within our high-throughput workflow using the modern PM6 Hamiltonian. Among other metrics, the ligand strain serves as an important quality indicator of the protein-ligand structures as it reflects both problematic ligand geometries typical in the conventional refinement and legitimate ligand deformation upon receptor binding. When compared with the originally deposited PDB structures, we found in all cases that QM refinement lead to ligand structures with much lower strain and on average, the strains were 3.5 fold lower. These results demonstrate the clear advantage of QM-augmented, high-throughput refinement.
This QM-based refinement tool has been joined with our MOE/DivCon product line in order to facilitate structure preparation and execution of the method. With this MOE plug-in installed, crystallographers and computational chemists are using these tools to better refine and re-refine structures in a much more rigorous manner to determine structure and discover new drug candidates!