Computational modelling of protein-ligand binding

The optimisation of protein-ligand interactions is a vital component in the various drug design strategies commonly employed today, including Structure Based Drug Design and Fragment Based Drug Design. One of the more sophisticated computational approaches to optimising the ligand to give improved binding affinity involves classical computer simulations combined with statistical mechanics allowing relative and absolute binding affinities to be calculated. However, such calculations are not without their difficulties. Specifically, the empirical force fields used are often inadequate to capture some of the more specific types of intermolecular interaction, and almost exclusively exclude explicit polarisation. If ligand binding is associated with changes in the number of explicitly bound waters in the protein-ligand complex, or results in significant reorganisation of the protein structure, then again current computational approaches will not reliably capture these processes.

To address these outstanding issues, the extension of classical molecular dynamics and Monte Carlo simulations is proposed. The use of empirical polarisable force fields or QM/MM methods may help address the deficiencies of existing force fields, while Grand Canonical Monte Carlo and other enhanced sampling approaches, including Hamiltonian Replica Exchange and Hybrid Monte Carlo methods, may assist in improving overall sampling and hence capture the large-scale structural reorganisations often associated with ligand binding events.

In this project, these sophisticated simulation approaches will be further developed and applied in the context of protein-ligand binding. Specific emphasis will be placed on the use of these methods on systems of current pharmaceutical interest, and collaboration with the pharma industry will be used to ensure the relevance and impact of the methodology developed.