A Generalized Framework Based on Enveloping Distribution Sampling for Calculating Geometrical Free-Energy Differences
- Preprint on ChemRxiv
Enveloping distribution sampling (EDS) is an efficient multistate free-energy method that has been mainly applied to estimate alchemical transformations, e.g., for relative solvation or binding free energies. In such simulations, each end-state in the definition of the EDS reference-state Hamiltonian represents a different solute or ligand. In contrast, when estimating geometrical (or conformational) free-energy differences, all end-states possess the same physical Hamiltonian. Thus, we found that the end-states have to be constructed with applied restraints (e.g. dihedral-angle or distance restraints) in such a way that they counteract the physical conformational minima to produce an effective flat potential along the reaction coordinate. The underlying theory and key aspects of the methodology are laid out and the proposed EDS framework is validated against umbrella sampling (US) on four test systems. Excellent agreement of the potentials of mean force is observed while requiring only a single EDS simulation in contrast to many parallel US windows.
This repository contains input files required to reproduce this work in the input_files folder and an example analysis in the analysis folder.
Domen Pregeljc (@dpregeljc)