Efficient solvation free energy simulations: impact of soft-core potential and a new adaptive λ-spacing method

Abstract

ABSTRACTThe calculation of Gibbs free energy of solvation ΔG is commonly applied to predict the solubility of solutes in different solvents. The transition path between the thermodynamic end states of full and non-existent solute–solvent interactions is described by the spacing of intermediate states and the soft-core potential employed to scale the interactions. The choice of both, the soft-core potential and the distribution of intermediate states, has a large impact on the simulated ΔG. In this work, we determine the free energy of hydration for 10 neutral amino acid side chain analogues to analyse the impact of different soft-core potentials on the consistency of results from two different free energy methods, i.e. thermodynamic integration and free energy perturbation. Additionally, we propose a new method to improve the alchemical pathway by adaptive spacing of intermediate states. Our technique is easy to implement and not limited to any simulation package or free energy technique in particular, making it easily adaptable to a variety of workflows. Exploiting the ability to systematically optimise alchemical pathways, we introduce a scheme to define a suitable number of windows based on prescribed configurational space overlap. With this, excessive sampling is avoided and computational costs can be reduced.