Computing hydration free energies of small molecules with first principles accuracy

J. Harry Moore,Daniel J. Cole,Gabor Csanyi

2024-06-22

Abstract:Free energies play a central role in characterising the behaviour of chemical systems and are among the most important quantities that can be calculated by molecular dynamics simulations. The free energy of hydration in particular is a well-studied physicochemical property of drug-like molecules and is commonly used to assess and optimise the accuracy of nonbonded parameters in empirical forcefields, and as a fast-to-compute surrogate of performance for protein-ligand binding free energy estimation. Machine learned potentials (MLPs) show great promise as more accurate alternatives to empirical forcefields, but are not readily decomposed into physically motivated functional forms, which has thus far rendered them incompatible with standard alchemical free energy methods that manipulate individual pairwise interaction terms. However, since the accuracy of free energy calculations is highly sensitive to the forcefield, this is a key area in which MLPs have the potential to address the shortcomings of empirical forcefields. In this work, we introduce an efficient alchemical free energy method compatible with MLPs, enabling, for the first time, calculations of biomolecular free energy with \textit{ab initio} accuracy. Using a pretrained, transferrable, alchemically equipped MACE model, we demonstrate sub-chemical accuracy for the hydration free energies of organic molecules.

Chemical Physics

What problem does this paper attempt to address?

The paper aims to address the problem of calculating the hydration free energy of small molecules, particularly in the context of drug discovery. Specifically, it introduces a method for calculating hydration free energy with accuracy comparable to ab initio methods through the use of machine learning potentials (MLPs). ### Main Problems Addressed by the Paper: 1. **Improving the Accuracy of Hydration Free Energy Calculations**: - Hydration free energy is one of the important physicochemical properties for evaluating drug molecules and is often used to assess and optimize the accuracy of non-bonded parameters in empirical force fields. - Traditional empirical force fields have shortcomings in describing certain interactions, leading to inaccurate hydration free energy calculations. 2. **Developing Free Energy Methods Compatible with Machine Learning Potentials**: - Machine learning potentials (MLPs) offer higher accuracy but cannot be directly decomposed into physically motivated forms, making them incompatible with standard variational free energy methods. - This paper proposes a new variational free energy method that is compatible with machine learning potentials, enabling high-accuracy biomolecular free energy difference calculations. 3. **Achieving Fast Convergence and Sub-Chemical Accuracy Calculations**: - By using pre-trained, transferable, variationally functional MACE models, the paper demonstrates sub-chemical accuracy in the calculation of organic molecule hydration free energy. - This method not only improves calculation accuracy but also achieves fast convergence, allowing free energy simulations to be completed within a reasonable timeframe. In summary, the paper addresses the shortcomings of traditional empirical force fields in hydration free energy calculations by introducing a new variational free energy method and utilizing machine learning potentials to achieve higher accuracy and faster convergence. This achievement is of significant importance for molecular simulations in drug discovery.

Computing hydration free energies of small molecules with first principles accuracy

Alchemical prediction of hydration free energies for SAMPL

Enhancing Accuracy and Feature Insights in Hydration Free Energy Predictions for Small Molecules with Machine Learning

Physics-based Machine Learning to Predict Hydration Free Energies for Small Molecules with a minimal number of descriptors: Interpretable and Accurate

Towards chemical accuracy for alchemical free energy calculations with hybrid physics-based machine learning / molecular mechanics potentials

Accurate Free Energy Calculation via Multiscale Simulations Driven by Hybrid Machine Learning and Molecular Mechanics Potentials

Predicting hydration free energies of the FreeSolv database of druglike molecules with molecular density functional theory

Exploring the Limits of the Generalized CHARMM and AMBER Force Fields through Predictions of Hydration Free Energy of Small Molecules

Comparative Assessment of Computational Methods for Free Energy Calculations of Ionic Hydration

Predicting hydration energies for multivalent ions

Predicting Organic Crystal Lattice Energies with Chemical Accuracy

Fast Computation of Solvation Free Energies with Molecular Density Functional Theory: Thermodynamic-Ensemble Partial Molar Volume Corrections

Prediction of hydration free energies for aliphatic and aromatic chloro derivatives using molecular dynamics simulations with the OPLS-AA force field

Teaching free energy calculations to learn from experimental data

Hydration Free Energies of Polypeptides from Popular Implicit Solvent Models versus All-Atom Simulation Results Based on Molecular Quasichemical Theory

Prediction of hydration free energies for the SAMPL4 diverse set of compounds using molecular dynamics simulations with the OPLS-AA force field

First Principles Free-Energy Theory of Solvation with Atomic Scale Liquid Structure

Fast calculation of hydrogen-bond strengths and free energy of hydration of small molecules

Calculated hydration free energies become less accurate with increases in molecular weight

Predicting solvation free energies with an implicit solvent machine learning potential

Computing spatially resolved rotational hydration entropies from atomistic simulations