Evaluating and improving the description of London dispersion interactions in molecular mechanical force fields using the exchange-hole dipole moment model

Mohebifar, Mohamad (2018) Evaluating and improving the description of London dispersion interactions in molecular mechanical force fields using the exchange-hole dipole moment model. Masters thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (7MB)

Abstract

Molecular simulations are used extensively to model processes in biophysics and biochemistry. These methods approximate the intramolecular and intermolecular interactions of the molecules in the system with a set of simplified mathematical expressions. London dispersion forces account for a significant portion of intermolecular interactions. These interactions play an important role in condensed matter physics and many biophysical phenomena. In this thesis, the eXchange-hole Dipole Moment model (XDM) of density functional theory was used to evaluate the dispersion coefficients in popular molecular mechanical models that are often used for simulations of water, organic molecules, and proteins. The dispersion coefficients derived from XDM calculations were compared to those extracted from molecular mechanical models with parameters from the GAFF, CGenFF, and OPLS force fields. For the generalized force fields, 88 organic molecules were evaluated. The Amber ff14sb, OPLS-AA, and CHARMM36 protein force fields were also evaluated using side chains models. Generally, the force field molecular C₆ dispersion coefficients overestimate the XDM C₆ dispersion coefficients by 50{60%. Despite this, these models predict the solvation energies of these molecules correctly. This trend was attributed to the neglect of higher order dispersion terms. In the empirical parameterization of these force fields, the interaction energy that should arise from these higher order terms will be spuriously added to the C₆ term. In the final chapter, a water model was developed with an improved non-bonded potential that describes repulsive forces more accurately using an exponential Buckingham-type term and includes C₆ and C₈ dispersion terms. High-performance GPU-CUDA and vectorized expressions for this potential were implemented in OpenMM. The model is able to predict the structural, physical, and transport properties of liquid water accurately.

Actions (login required)

View Item