Since the majority of biochemical processes occur in the solution-phase, understanding the solvation behaviour of biomolecules and their aggregates is a fundamental scientific problem that is also important in many practical applications such as pharmaceutical research and development.   We are developing chemically accurate methods for modelling biomolecules in solution based on the Integral Equation Theory (IET) of Molecular Liquids.   We have recently applied these methods to compute hydration free energy, solubility of crystalline organic molecules, and protein-ligand binding free energies.

Bovine chymosin is a mammalian aspartic protease found in the fourth stomach of calves, where it aids digestion by selectively cleaving the milk protein k-casein. The enzyme is important in the food industry because it has been sold to initiate milk clotting in the manufacturing of processed dairy products since 1874.  Although bovine chymosin remains in common use, it has recently been demonstrated that the camel variant of the enzyme has significantly higher clotting activity and much lower unspecific protease activity for bovine milk, which has led to it being successfully marketed as an alternative to the bovine enzyme. By contrast, bovine chymosin has a very low catalytic rate for proteolysis of camel k-casein.   The difference in catalytic efficacy is not well understood on a molecular level due in part to a lack of structural information about the chymosin-k-casein complexes.  We are using molecular simulation and free energy calculations to shed light on the observed disparity in catalytic function and to suggest small modifications of the existing enzymes that would improve their function.

Proteins exhibit internal dynamics on a wide range of time-scales from bond stretching vibrations involving hydrogen to large-scale conformational changes involving whole domain movements.  The slower, large-scale internal dynamics are important for the biological function of proteins, but they are difficult to study using standard computational methods, which do not permit a full sampling of these conformational degrees of freedom.  For example, in regular molecular dynamics (MD) simulations using atomistic force fields, the forces are used for guiding the sampling, and this has the consequence that a simulation spends significant time sampling a limited region of phase space.  We are developing hybrid atomistic/coarse-grained simulation methods that allow enhanced sampling of the dynamics of biomacromolecules.

Understanding how molecules recognize each other in biological environments is one of the fundamental issues in the biomolecular sciences. It is also importance in many areas of industrial research including the pharmaceutical industry, where predictions of the binding modes of small molecules (e.g. waters, ions, drugs) to biological macromolecules (e.g. proteins, DNA) are used to help design new pharmaceuticals. Since solvent plays an important role in mediating many molecular recognition events in biological systems, it is vital that computational methods to model biomolecular recognition incorporate an adequate molecular-scale model of solvent.  We are combining molecular theories of solvation with high-performance computing techniques to develop new methods to compute binding free energies and binding modes of biological complexes.

Our research is driven by a scientific curiousity that has often taken us off happily in new directions.   For example, we have recently been interested in how to validate phosphorylation sites in proteins and whether molecular informatics can be used to reduce doping in sport.  We are always interested to discuss possible research collaborations with scientists of any background.