Rapid Prediction of Solvation Free Energy. 2. The First-Shell Hydration (FiSH) Continuum Model

Abstract

Local ordering of water in the first hydration shell around a solute is different from isotropic bulk water. This leads to effects that are captured by explicit solvation models and missed by continuum solvation models which replace the explicit waters with a continuous medium. In this paper, we introduce the First-Shell Hydration (FiSH) model as a first attempt to introduce first-shell effects within a continuum solvation framework. One such effect is charge asymmetry, which is captured by a modified electrostatic term within the FiSH model by introducing a nonlinear correction of atomic Born radii based on the induced surface charge density. A hybrid van der Waals formulation consisting of two continuum zones has been implemented. A shell of water restricted to and uniformly distributed over the solvent-accessible surface (SAS) represents the first solvation shell. A second region starting one solvent diameter away from the SAS is treated as bulk water with a uniform density function. Both the electrostatic and van der Waals terms of the FiSH model have been calibrated against linear interaction energy (LIE) data from molecular dynamics simulations. Extensive testing of the FiSH model was carried out on large hydration data sets including both simple compounds and drug-like molecules. The FiSH model accurately reproduces contributing terms, absolute predictions relative to experimental hydration free energies, and functional class trends of LIE MD simulations. Overall, the implementation of the FiSH model achieves a very acceptable performance and transferability improving over previously developed solvation models, while being complemented by a sound physical foundation.

Rapid Prediction of Solvation Free Energy. 1. An Extensive Test of Linear Interaction Energy (LIE)

“…This data will then be used to train the respective components of purely continuum solvation models with the aim of having this model mimic an explicit solvent simulation (see accompanying paper 35 ).…”

Rapid prediction of solvation free energy. 3. Application to the SAMPL2 challenge

“…The FiSH continuum solvation model has been developed to mimic the accuracy of an explicit-solvent LIE hydration model [34]. This continuum model has been trained on hydration data from explicit-solvent simulations rather than on experimental hydration data in order to enable the calibration of its electrostatic and non-polar components on the corresponding terms available from explicit-solvent simulations but not accessible experimentally.…”

“…Furthermore, the LIE approach was favoured over FEP-like methods for the calibration of this continuum model due to its term decomposition that is simpler and more compatible with the solvation contributions typically calculated with continuum models, and its similar accuracy in predicting hydration free energies relative to the more expensive FEP approach. A complete report of the FiSH continuum model has been described elsewhere [34] and only a short outline will follow. The FiSH continuum model includes an electrostatic, a van der Waals and a cavity term similar to the explicitsolvent LIE hydration model it was trained and tested on,…”

“…The Born radii are derived from the GAFF van der Waals radii with a correction based on the induced surface charge density (ISCD) on the molecular surface [34]. The radius correction is designed to capture the charge asymmetry in the response of water to positively and negatively charged groups [45][46][47][48][49][50].…”

“…We have recently revisited the SAMPL1 set with both an explicit-solvent model using the LIE approximation [33] and with FiSH (First Shell Hydration), our refined continuum solvation model [34]. Both these models showed an increase in accuracy over our previous CED model.…”