Combining Machine Learning and Enhanced Sampling Techniques for Efficient and Accurate Calculation of Absolute Binding Free Energies.

Rhys Evans,Benjamin P Cossins,Francesco L Gervasio,Carolina Estarellas,Ladislav Hovan,Gareth A Tribello

doi:10.1021/acs.jctc.0c00075

Abstract

Calculating absolute binding free energies is challenging and important. In this paper, we test some recently developed metadynamics-based methods and develop a new combination with a Hamiltonian replica-exchange approach. The methods were tested on 18 chemically diverse ligands with a wide range of different binding affinities to a complex target; namely, human soluble epoxide hydrolase. The results suggest that metadynamics with a funnel-shaped restraint can be used to calculate, in a computationally affordable and relatively accurate way, the absolute binding free energy for small fragments. When used in combination with an optimal pathlike variable obtained using machine learning or with the Hamiltonian replica-exchange algorithm SWISH, this method can achieve reasonably accurate results for increasingly complex ligands, with a good balance of computational cost and speed. An additional benefit of using the combination of metadynamics and SWISH is that it also provides useful information about the role of water in the binding mechanism.

Highlights

Estimating target-ligand binding free energies (BFEs) is a challenging and important task in computer-aided drug discovery (CADD)
Recent advances in free energy perturbation (FEP) methodologies have made them amenable for routine and successful use in drug discovery pipelines.[10−13] this mainly applies to the determination of relative BFEs, which can be used in the hit-to-lead optimization phase, significant progress[14−16] has been made in the calculation of absolute binding free energies (ABFEs) using alchemical approaches, such as double decoupling methods.[17−23] the routine use of alchemical methods for the calculation of ABFEs still faces a number of challenges, especially with targets that undergo significant conformational changes, as well as with charged or noncongeneric ligands.[24−26] A valid alternative for performing ABFE calculations is found in collective-variable-based free energy methods
With generic collective variables (CVs) that can be used for a range of systems and by lowering the computational cost, we address the most pressing challenges hindering the adoption of CVbased free energy methods for ABFE evaluation in routine computer-aided drug discovery pipelines

Summary

Introduction

Estimating target-ligand binding free energies (BFEs) is a challenging and important task in computer-aided drug discovery (CADD). Umbrella sampling[27,28] and metadynamics[6,9,29] have repeatedly been used to compute the ABFE along physical binding trajectories associated with both simple and complex systems.[18,30−33] In contrast to alchemical ones, these methods can be used to directly enhance the exploration of target conformational changes. They explore metastable minima and transition states that determine binding kinetics while, due to their nature, alchemical methods only sample the bound and unbound states. Metadynamics and umbrella sampling have been combined with multiple replica approaches such as parallel tempering to improve their convergence with nonoptimal CVs.[36−38] These

Objectives

Methods

Results

Conclusion