News Item

Event: QuantumBio to present at American Crystallographic Association Conference - New Orleans, LA

May 28, 2017

QuantumBio staff to present at the 67th Annual Meeting of the American Crystallographic Association (ACA) in New Orleans, LA!

Title: Modern, rigorous macromolecular crystallographic refinement using mixed-QM/MM functional methods as implemented in DivCon
Speaker: 
Oleg Y. Borbulevych
Session:
General Interest I (Lecture 2.2.3)
Date/Time: 
16:45-17:00PM on Sunday, May 28, 2017

Abstract:

Conventional macromolecular crystallographic refinement relies on highly approximate functionals which lack explicit, chemically rigorous terms for electrostatics, polarization, dispersion, hydrogen bonds, and other interactions, and they often rely on pre-determined parameters to capture an a priori understanding of the structure. Previously we addressed this problem, especially for active sites and ligands, through the integration of our DivCon, linear scaling, semiempricial quantum mechanics software with the PHENIX package. This implementation was limited in that it still treated the remainder of the system using conventional, stereochemical restraints without taking into account, long- range interactions between this part and the QM region. In order to address this deficiency and capture a more complete understanding of the structure, we have developed a fully automated approach for macromolecular refinement based on a two layer, QM/MM (ONIOM) scheme implemented within our DivCon plugin. This method consists of one or more user-defined, PM6 quantum mechanics regions (referred to as the "region layer") coupled with a "system layer" encompassing the rest of a protein described with our implementation of MM functional. This chemically complete approach is fully automated for amino acids, nucleotides, truncated residues, ligands, cofactors, and so on, and it does not require user-provided stereochemical restraint files (CIF's) or other description files in order to correctly type the atoms within the structure. We have validated this plugin on a population of 80 protein-ligand structures from the Astex Diverse Set. Across the entire population, as with the previous Region-DivCon:QM implementation, the new DivCon:QM/MM plugin results in significantly lower strain energies and lower residual difference density around each ligand. Further, since QM/MM is being applied to the entire structure, the tool leads to dramatic improvements of the overall structural quality. For example, the MolProbity ClashScore averaged over 80 structures is 1.11 after DivCon:QM/MM refinement compared to 4.83 after conventional PHENIX refinement and to 5.52 after use of our previous, pure DivCon:QM region refinement.