Research Fellows Directory
Dr Ying-Chih Chiang
University of Southampton
As drug-resistant bacteria are causing extra medical expenses and productivity losses of up to 1.5 billion Euro per year and are posing serious risks to healthcare systems worldwide, I am interested in how computer simulations can help to design new antibiotics. Computer simulations rooted in quantitative physical laws can be used to accurately evaluate the activity of a drug, e.g. by predicting the binding free energy of an antibiotic to its receptor Penicillin Binding Protein (PBP). Through scanning the binding free energy between various antibiotics with the drug-susceptible and with the drug-resistant PBP, we can deduct rules of how to design a better antibiotic against drug-resistant bacteria. Although this idea is rather straightforward, its practical application is hampered because free energy calculations are computationally demanding. Such calculations employ molecular dynamics (MD) simulations to provide samples of the statistically relevant conformations, but rare events such as the protein conformational change upon drug binding cannot be sampled efficiently using traditional plain MD simulations, a problem which is known as the sampling inefficiency.
To deal with this sampling inefficiency, many enhanced sampling methods have been proposed but the current state of the art techniques, e.g. using GPU computing power to perform replica exchange with solute tempering (REST) simulations, are still too computationally demanding for our goal. Therefore, one must investigate the origin of the sampling inefficiency that occurs in the free energy calculations. It turns out that an important process, namely the relaxation process, can occur upon protein-drug binding, but the traditional equilibrium statistical mechanics cannot describe the free energy change during the relaxation. Hence, my current research focuses on utilizing non-equilibrium statistical mechanics to describe the free energy change due to the relaxation processes which can be sampled using methods inspired by quantum dynamics involving multiple electronic states. We expect that this approach can further reduce the computational effort for computer-aided drug design, and will help in the fight against drug-resistant PBP as well.
Interests and expertise (Subject groups)