Binding free energy and ligand orientation calculations using a Monte Carlo method with Markov Sate analysis

Abstract

Computing binding free energies has great implications in drug design. Using PELE technique, it has been shown that one can get quick and accurate estimations by means of a Markov state model3. We improved our methodology to compute faster binding relative free energy differences, mainly by analysis reducing the sampled region. This possibility opens a way in all-atom drug lead optimization by efficiently scoring a list of potential candidates in terms of binding affinities (approximately in 24hours), while still modeling accurately the protein-drug induced fit. Furthermore, we added information of the ligand orientation allowing us to obtain a better insight of the entrance mechanism. First, we show benchmark results - a series of benzamidine-like inhibitors in trypsin. Then, we apply our method to a more realistic scenario: the binding to a glucocorticoid receptor, and we show the performance for a new benchmark with a larger range of binding free energies (~14 kcal/mol). Simulations are obtained with our new in-house code PELE++, an improvement over the technique presented in references [1,2], (paper in preparation).