Bind3P: Optimization of a Water Model Based on Host–Guest Binding Data

Abstract

We report a water model, Bind3P (Version 0.1), which was obtained by using sensitivity analysis to readjust the Lennard-Jones parameters of the TIP3P model against experimental binding free energies for six host-guest systems, along with pure liquid properties. Tests of Bind3P against >100 experimental binding free energies and enthalpies for host-guest systems distinct from the training set show a consistent drop in the mean signed error, relative to matched calculations with TIP3P. Importantly, Bind3P also yields some improvement in the hydration free energies of small organic molecules and preserves the accuracy of bulk water properties, such as density and the heat of vaporization. The same approach can be applied to more sophisticated water models that can better represent pure water properties. These results lend further support to the concept of integrating host-guest binding data into force field parametrization.

Overview of the SAMPL6 host-guest binding affinity prediction challenge

“…Lennard-Jones interactions [101]. In addition, hosts in both families have been shown to bind ions that 126 can compete with and lower the binding affinity of other guests in solution [37,98,109]. Depending on 127 differences in concentration and composition, the effect on the binding free energy can be between 1-128 2 kcal/mol [85,94,98].…”

“…662 Linear corrections fit to prior experimental measurements do not improve predictive utility 663 The experimental-based correction adopted by several groups introduces a new theme in the challenge 664 which pertains to strategies that can be used to inject previous knowledge into molecular simulations. Force 665 field parameters are in principle capable of incorporating experimental data, but an update of the model 666 driven by binding free energy measurements or other ensemble observables is doubtlessly challenging and 667 may involve calculations as expensive as the production calculations so this is normally not routinely viable, 668 although previous studies indicated the validity and feasibility of such an approach [125,126]. Other schemes 669 that emerged in particular from the field of crystallographic structural refinement avoid modifying the force 670 field parameters and instead add one or more biasing terms to the simulation to replicate experimental 671 measurements that the underlying force field cannot reproduce [16,123].…”

Computational Prediction of Mutational Effects on the SARS-CoV-2 Binding by Relative Free Energy Calculations

“…First, MD simulations were necessary to relax the undesired contacts both in the homology model and in the crystal structure, and thus the quality of the protein force field was heavily relied on to produce conformations close to the native states. A more accurate general force field and water model could be developed and used to improve on predicting binding properties [48][49][50] . Secondly, our single-point alaninescanning calculations only yielded the contribution of the individual residue without taking the coupling effect caused by double or multiple mutations into account.…”

SAMPL7 Host-Guest Challenge Overview: Assessing the reliability of polarizable and non-polarizable methods for binding free energy calculations

“…Full performance statistics are in Table 2. model, and its combination with explicit fixed charge water models typically results in predicted free energies that are too favorable (particularly, prior work has shown that GAFF with TIP3P leads to a consistent error in this direction for guests containing carboxylates and alcohols [49]). This is exactly the case here for systems with guests containing a carboxylate for the…”