Developing enhanced sampling methods for modelling hydration in Protein-Ligand Systems

Abstract

Molecular dynamics (MD) is a widely used computational technique for gaininginsight into biomolecular systems and for providing direction for further experimental research, particularly in the context of drug discovery. One of the drawbacks of MD is its difficulty in sampling slow degrees of freedom that occur on the microsecond timescale and longer, given the currently available computational hardware. As a result, the development of enhanced sampling methods, whereby the kinetic barriers preventing good sampling are bypassed, is of significant importance for improving the application of MD to pharmaceutical research.To this end, this thesis presents the implementation, validation and combination of a number of different enhanced sampling methods, primarily focused on improving the sampling of water molecules bound in the binding site of a protein.Initially, nonequilibrium sampling is used to improve the acceptance rates ofinstantaneous grand canonical Monte Carlo moves for the sampling of these bound waters. This results in not only an increase in efficiency, but also an improvement in the conformational sampling of the protein and ligand. An existing Python library, OpenMMSLICER, is then adapted to allow for the direct calculation of relative binding free energies (RBFEs) in both an equilibrium and nonequilibrium manner. The implementation is validated by comparing estimated hydration free energies of a series of small molecules to experimental data.The focus then turns to the combination of these two methods, as the first application of nonequilibrium grand canonical sampling to overcome the trapped water problem in relative binding free energy calculations is presented. Results demonstrate that this new technique improves the accuracy of the relative affinities when compared to experimental data, in the case of both equilibrium and nonequilibrium RBFE calculations. Finally, the application of many of the methods introduced over the course of this work are combined on a test case of current pharmaceutical interest.

More information

Identifiers

ORCID for Jonathan Essex: orcid.org/0000-0003-2639-2746

Catalogue record

Export record

Altmetrics