| Summary: | The accurate prediction of the binding affinities of ligands to proteins is a major goal in drug discovery and personalised medicine. The use of in silico methods to predict binding affinities has been largely confined to academic research until recently, primarily due to the lack of their reproducibility, as well as unaffordably longer time to solution. In this thesis, I mainly describe the ensemble based molecular dynamics approaches, ESMACS and TIES, that provide a route to reliable predictions of free energies meeting the requirements of speed, accuracy, precision and reliability. The performance of both these methods when applied to a diverse set of protein targets and ligands is reported. The results are in very good agreement with experimental data while the methods are repeatable by construction. Statistical uncertainties of the order of 0.5 kcal/mol or less are achieved. These methods have been further extended to incorporate enhanced sampling techniques based on replica exchange (also known as parallel tempering) to handle situations where conformational sampling is difficult using standard molecular dynamics. A critical assessment of free energy estimators like MBAR has been made for their application in binding affinity prediction. The methodologies described are shown to have a positive impact in the drug design process in the pharmaceutical domain as well as in personalised medicine, with concomitant potential major industrial and societal impact. Finally, our automated workflow, comprising the Binding Affinity Calculator (BAC) together with the FabSim are described. These tools and services help us complete the entire execution in 8 hours or less, depending on the high performance architecture and hardware available.
|
|---|
