Implicit Solvents for the Polarizable Atomic Multipole AMOEBA Force Field

Abstract

Computational protein design, ab initio protein/RNA folding, and protein-ligand screening can be too computationally demanding for explicit treatment of solvent. For these applications, implicit solvent offers a compelling alternative, which we describe here for the polarizable atomic multipole AMOEBA force field based on three treatments of continuum electrostatics: numerical solutions to the Poisson-Boltzmann equation (PBE), the domain-decomposition Conductor-like Screening Model (ddCOSMO) approximation to the PBE, and the analytic generalized Kirkwood (GK) approximation. The continuum electrostatic models are combined with a nonpolar estimator based on novel cavitation and dispersion terms. Electrostatic model parameters are numerically optimized using a least squares style target function based on a library of 103 small molecule solvation free energy differences. Mean signed errors for the APBS, ddCOSMO, and GK models are 0.05, 0.00, and 0.00 kcal/mol, respectively, while the mean unsigned errors are 0.70, 0.63, and 0.51 kcal/mol, respectively. Validation of the electrostatic response of the resulting implicit solvents, which are available in the Tinker (or Tinker-HP), OpenMM, and Force Field X software packages, is based on comparisons to explicit solvent simulations for a series of proteins and nucleic acids. Overall, the emergence of performative implicit solvent models for polarizable force fields will open the door to their use for folding and design applications.<br>

CHARMM-GUI Implicit Solvent Modeler for Various Generalized Born Models in Different Simulation Programs

“…71 No specific RMSD value was reported for 1TSU by Nguyen et al 71 f 2HKB and 1MIS systems were used for GB-HTC using the AMOEBA FF. 84 ΔG solv , the solvation free energy, was calculated based on eq 1. The nonpolar energy was estimated using the solventaccessible surface area (SASA), while the polar contribution was calculated using GB.…”

Protein Structure Prediction Using a Maximum Likelihood Formulation of a Recurrent Geometric Network

“…Traditional computational methods for predicting a protein fold combine a physics-based model of intermolecular forces (8)(9)(10)(11)(12)(13)(14)(15)(16), an explicit (17) or continuum (8)(9)(10)(11)(12)(13)(14)(15)(16)(18)(19)(20) solvation model, and a sampling algorithm (5,21) that builds upon molecular dynamics simulations. However, a limiting factor of a physics-based approaches is that protein folding often occurs on millisecond (10 -3 seconds) or longer timescales, whereas GPU-accelerated molecular dynamics simulations are largely limited to microsecond (10 -6 seconds) time scales due to the computational expense of computing interactions over all atoms in a protein system.…”