Parameterization of a reactive force field using a Monte Carlo algorithm

Parameterization of a molecular dynamics force field is essential in realistically modeling the physicochemical processes involved in a molecular system. This step is often challenging when the equations involved in describing the force field are complicated as well as when the parameters are mostly empirical. ReaxFF is one such reactive force field which uses hundreds of parameters to describe the interactions between atoms. The optimization of the parameters in ReaxFF is done such that the properties predicted by ReaxFF matches with a set of quantum chemical or experimental data. Usually, the optimization of the parameters is done by an inefficient single‐parameter parabolic‐search algorithm. In this study, we use a robust metropolis Monte‐Carlo algorithm with simulated annealing to search for the optimum parameters for the ReaxFF force field in a high‐dimensional parameter space. The optimization is done against a set of quantum chemical data for MgSO₄ hydrates. The optimized force field reproduced the chemical structures, the equations of state, and the water binding curves of MgSO₄ hydrates. The transferability test of the ReaxFF force field shows the extend of transferability for a particular molecular system. This study points out that the ReaxFF force field is not indefinitely transferable. © 2013 Wiley Periodicals, Inc.     

Global optimization of parameters in the reactive force field ReaxFF for SiOH

We have used unbiased global optimization to fit a reactive force field to a given set of reference data. Specifically, we have employed genetic algorithms (GA) to fit ReaxFF to SiOH data, using an in‐house GA code that is parallelized across reference data items via the message‐passing interface (MPI). Details of GA tuning turn‐ed out to be far less important for global optimization efficiency than using suitable ranges within which the parameters are varied. To establish these ranges, either prior knowledge can be used or successive stages of GA optimizations, each building upon the best parameter vectors and ranges found in the previous stage. We have finally arrive‐ed at optimized force fields with smaller error measures than those published previously. Hence, this optimization approach will contribute to converting force‐field fitting from a specialist task to an everyday commodity, even for the more difficult case of reactive force fields. © 2013 Wiley Periodicals, Inc.     

Modeling the effect of ion-induced shock waves and DNA breakage with the reactive CHARMM force field

Ion-induced DNA damage is an important effect underlying ion beam cancer therapy. This article introduces the methodology of modeling DNA damage induced by a shock wave caused by a projectile ion. Specifically it is demonstrated how single- and double strand breaks in a DNA molecule could be described by the reactive CHARMM (rCHARMM) force field implemented in the program MBN Explorer. The entire workflow of performing the shock wave simulations, including obtaining the crucial simulation parameters, is described in seven steps. Two exemplary analyses are provided for a case study simulation serving to: (a) quantify the shock wave propagation and (b) describe the dynamics of formation of DNA breaks. The article concludes by discussing the computational cost of the simulations and revealing the possible maximal computational time for different simulation set-ups.     

A new GROMOS force field for hexopyranose-based carbohydrates

A new parameter set (referred to as 45A4) is developed for the explicit-solvent simulation of hexopyranose-based carbohydrates. This set is compatible with the most recent version of the GROMOS force field for proteins, nucleic acids, and lipids, and the SPC water model. The parametrization procedure relies on: (1) reassigning the atomic partial charges based on a fit to the quantum-mechanical electrostatic potential around a trisaccharide; (2) refining the torsional potential parameters associated with the rotations of the hydroxymethyl, hydroxyl, and anomeric alkoxy groups by fitting to corresponding quantum-mechanical profiles for hexopyranosides; (3) adapting the torsional potential parameters determining the ring conformation so as to stabilize the (experimentally predominant) (4)C(1) chair conformation. The other (van der Waals and nontorsional covalent) parameters and the rules for third and excluded neighbors are taken directly from the most recent version of the GROMOS force field (except for one additional exclusion). The new set is general enough to define parameters for any (unbranched) hexopyranose-based mono-, di-, oligo- or polysaccharide. In the present article, this force field is validated for a limited set of monosaccharides (alpha- and beta-D-glucose, alpha- and beta-D-galactose) and disaccharides (trehalose, maltose, and cellobiose) in solution, by comparing the results of simulations to available experimental data. More extensive validation will be the scope of a forthcoming article. (c) 2005 Wiley Periodicals, Inc. J Comput Chem 26: 1400-1412, 2005.     

Paramfit: Automated optimization of force field parameters for molecular dynamics simulations

The generation of bond, angle, and torsion parameters for classical molecular dynamics force fields typically requires fitting parameters such that classical properties such as energies and gradients match precalculated quantum data for structures that scan the value of interest. We present a program, Paramfit, distributed as part of the AmberTools software package that automates and extends this fitting process, allowing for simplified parameter generation for applications ranging from single molecules to entire force fields. Paramfit implements a novel combination of a genetic and simplex algorithm to find the optimal set of parameters that replicate either quantum energy or force data. The program allows for the derivation of multiple parameters simultaneously using significantly fewer quantum calculations than previous methods, and can also fit parameters across multiple molecules with applications to force field development. Paramfit has been applied successfully to systems with a sparse number of structures, and has already proven crucial in the development of the Assisted Model Building with Energy Refinement Lipid14 force field. © 2014 Wiley Periodicals, Inc.     

Development of a force field for Li(2)SiF(6)

A force field has been developed for Li(2)SiF(6) for subsequent use in Molecular Dynamics (MD) simulations involving Li(+) and SiF(2-) (6) ions in a polymer electrolyte host. Both ab initio calculations and available empirical data have been used. The force field has been verified in simulations of the crystal structure of Li(2)SiF(6) in two different space groups: P321 and P3(-)m1. The use of MD simulation to assess the correct space group for Li(2)SiF(6) shows that it is probably P321.     

An improved nucleic acid parameter set for the GROMOS force field

Over the past decades, the GROMOS force field for biomolecular simulation has primarily been developed for performing molecular dynamics (MD) simulations of polypeptides and, to a lesser extent, sugars. When applied to DNA, the 43A1 and 45A3 parameter sets of the years 1996 and 2001 produced rather flexible double-helical structures, in which the Watson-Crick hydrogen-bonding content was more limited than expected. To improve on the currently available parameter sets, the nucleotide backbone torsional-angle parameters and the charge distribution of the nucleotide bases are reconsidered based on quantum-chemical data. The new 45A4 parameter set resulting from this refinement appears to perform well in terms of reproducing solution NMR data and canonical hydrogen bonding. The deviation between simulated and experimental observables is now of the same order of magnitude as the uncertainty in the experimental values themselves.     

CHARMM force field parameters for simulation of reactive intermediates in native and thio-substituted ribozymes

Force field parameters specifically optimized for residues important in the study of RNA catalysis are derived from density-functional calculations, in a fashion consistent with the CHARMM27 all-atom empirical force field. Parameters are presented for residues that model reactive RNA intermediates and transition state analogs, thio-substituted phosphates and phosphoranes, and bound Mg(2+) and di-metal bridge complexes. Target data was generated via density-functional calculations at the B3LYP/6-311++G(3df,2p)// B3LYP/6-31++G(d,p) level. Partial atomic charges were initially derived from CHelpG electrostatic potential fitting and subsequently adjusted to be consistent with the CHARMM27 charges. Lennard-Jones parameters were determined to reproduce interaction energies with water molecules. Bond, angle, and torsion parameters were derived from the density-functional calculations and renormalized to maintain compatibility with the existing CHARMM27 parameters for standard residues. The extension of the CHARMM27 force field parameters for the nonstandard biological residues presented here will have considerable use in simulations of ribozymes, including the study of freeze-trapped catalytic intermediates, metal ion binding and occupation, and thio effects.     

A systematic approach to calibrate a transferable polarizable force field parameter set for primary alcohols

In this work, parameters are optimized for a charge‐on‐spring based polarizable force field for linear alcohols. We show that parameter transferability can be obtained using a systematic approach in which the effects of parameter changes on physico‐chemical properties calculated from simulation are predicted. Our previously described QM/MM calculations are used to attribute condensed‐phase polarizabilities, and starting from the non‐polarizable GROMOS 53A5/53A6 parameter set, van der Waals and Coulomb interaction parameters are optimized to reproduce pure‐liquid (thermodynamic, dielectric, and transport) properties, as well as hydration free energies. For a large set of models, which were obtained by combining small perturbations of 10 distinct parameters, values for pure‐liquid properties of the series methanol to butanol were close to experiment. From this large set of models, we selected 34 models without special repulsive van der Waals parameters to distinguish between hydrogen‐bonding and non‐hydrogen‐bonding atom pairs, to make the force field simple and transparent. © 2017 Wiley Periodicals, Inc.     

Development of the force field parameters for phosphoimidazole and phosphohistidine

Phosphorylation of histidine-containing proteins is a key step in the mechanism of many phosphate transfer enzymes (kinases, phosphatases) and is the first stage in a wide variety of signal transduction cascades in bacteria, yeast, higher plants, and mammals. Studies of structural and dynamical aspects of such enzymes in the phosphorylated intermediate states are important for understanding the intimate molecular mechanisms of their functioning. Such information may be obtained via molecular dynamics and/or docking simulations, but in this case appropriate force field parameters for phosphohistidine should be explicitly defined. In the present article we describe development of the GROMOS96 force field parameters for phosphoimidazole molecule--a realistic model of the phosphohistidine side chain. The parameterization is based on the results of ab initio quantum chemical calculations with subsequent refinement and testing using molecular mechanics and molecular dynamics simulations. The set of force constants and equilibrium geometry is employed to derive force field for the phosphohistidine moiety. Resulting parameters and topology are incorporated into the molecular modeling package GROMACS and used in molecular dynamics simulations of a phosphohistidine-containing protein in explicit solvent.     

diSBPred: A machine learning based approach for disulfide bond prediction

The protein disulfide bond is a covalent bond that forms during post-translational modification by the oxidation of a pair of cysteines. In protein, the disulfide bond is the most frequent covalent link between amino acids after the peptide bond. It plays a significant role in three-dimensional (3D) ab initio protein structure prediction (aiPSP), stabilizing protein conformation, post-translational modification, and protein folding. In aiPSP, the location of disulfide bonds can strongly reduce the conformational space searching by imposing geometrical constraints. Existing experimental techniques for the determination of disulfide bonds are time-consuming and expensive. Thus, developing sequence-based computational methods for disulfide bond prediction becomes indispensable. This study proposed a stacking-based machine learning approach for disulfide bond prediction (diSBPred). Various useful sequence and structure-based features are extracted for effective training, including conservation profile, residue solvent accessibility, torsion angle flexibility, disorder probability, a sequential distance between cysteines, and more. The prediction of disulfide bonds is carried out in two stages: first, individual cysteines are predicted as either bonding or non-bonding; second, the cysteine-pairs are predicted as either bonding or non-bonding by including the results from cysteine bonding prediction as a feature.The examination of the relevance of the features employed in this study and the features utilized in the existing nearest neighbor algorithm (NNA) method shows that the features used in this study improve about 7.39 % in jackknife validation balanced accuracy. Moreover, for individual cysteine bonding prediction and cysteine-pair bonding prediction, diSBPred provides a 10-fold cross-validation balanced accuracy of 82.29 % and 94.20 %, respectively. Altogether, our predictor achieves an improvement of 43.25 % based on balanced accuracy compared to the existing NNA based approach. Thus, diSBPred can be utilized to annotate the cysteine bonding residues of protein sequences whose structures are unknown as well as improve the accuracy of the aiPSP method, which can further aid in experimental studies of the disulfide bond and structure determination.     

Benchmarking PES-Learn's machine learning models predicting accurate potential energy surface for quantum scattering

Machine learning (ML) models, neural networks, and Gaussian processes have been used to predict the potential energy surface taking C₂-He (both static and dynamic scenario) and NCCN-He collision systems. The surface is restricted to ∽125 points where traditional spline becomes inefficacious. Quantum dynamics is performed by solving close-coupling equation to compute cross sections benchmarking the performance of the ML models. The current study forms a basis for any future investigation of larger molecules where conventional fitting fails due to sparser ab initio points and cuts down the computational time without compromising on the quality of the surface.     

Polarizable empirical force field for sulfur‐containing compounds based on the classical Drude oscillator model

Condensed‐phase computational studies of molecules using molecular mechanics approaches require the use of force fields to describe the energetics of the systems as a function of structure. The advantage of polarizable force fields over nonpolarizable (or additive) models lies in their ability to vary their electronic distribution as a function of the environment. Toward development of a polarizable force field for biological molecules, parameters for a series of sulfur‐containing molecules are presented. Parameter optimization was performed to reproduce quantum mechanical and experimental data for gas phase properties including geometries, conformational energies, vibrational spectra, and dipole moments as well as for condensed phase properties such as heats of vaporization, molecular volumes, and free energies of hydration. Compounds in the training set include methanethiol, ethanethiol, propanethiol, ethyl methyl sulfide, and dimethyl disulfide. The molecular volumes and heats of vaporization are in good accordance with experimental values, with the polarizable model performing better than the CHARMM22 nonpolarizable force field. Improvements with the polarizable model were also obtained for molecular dipole moments and in the treatment of intermolecular interactions as a function of orientation, in part due to the presence of lone pairs and anisotropic atomic polarizability on the sulfur atoms. Significant advantage of the polarizable model was reflected in calculation of the dielectric constants, a property that CHARMM22 systematically underestimates. The ability of this polarizable model to accurately describe a range of gas and condensed phase properties paves the way for more accurate simulation studies of sulfur‐containing molecules including cysteine and methionine residues in proteins. © 2010 Wiley Periodicals, Inc. J Comput Chem, 2010     

Recognition of protein allosteric states and residues: Machine learning approaches

Allostery is a process by which proteins transmit the effect of perturbation at one site to a distal functional site upon certain perturbation. As an intrinsically global effect of protein dynamics, it is difficult to associate protein allostery with individual residues, hindering effective selection of key residues for mutagenesis studies. The machine learning models including decision tree (DT) and artificial neural network (ANN) models were applied to develop classification model for a cell signaling allosteric protein with two states showing extremely similar tertiary structures in both crystallographic structures and molecular dynamics simulations. Both DT and ANN models were developed with 75% and 80% of predicting accuracy, respectively. Good agreement between machine learning models and previous experimental as well as computational studies of the same protein validates this approach as an alternative way to analyze protein dynamics simulations and allostery. In addition, the difference of distributions of key features in two allosteric states also underlies the population shift hypothesis of dynamics‐driven allostery model. © 2018 Wiley Periodicals, Inc.     

Automatic parameterization of force fields for liquids by simplex optimization

In this study we demonstrate an automatic method of force field development for molecular simulations. Parameter tuning is taken as an optimization problem in many dimensions. The parameters are automatically adapted to reproduce known experimental data such as the density and the heat of vaporization. Our method is more systematic than guessing parameters and, at the same time, saves human labor in parameterization. It was applied successfully to several molecular liquids. As a test, force fields for 2-methylpentane, tetrahydrofurane, cyclohexene, and cyclohexane were developed. ©1999 John Wiley & Sons, Inc. J Comput Chem 20: 1009–1017, 1999     

Development and application of a ReaxFF reactive force field for hydrogen combustion

To investigate the reaction kinetics of hydrogen combustion at high-pressure and high-temperature conditions, we constructed a ReaxFF training set to include reaction energies and transition states relevant to hydrogen combustion and optimized the ReaxFF force field parameters against training data obtained from quantum mechanical calculations and experimental values. The optimized ReaxFF potential functions were used to run NVT MD (i.e., molecular dynamics simulation with fixed number of atoms, volume, and temperature) simulations for various H(2)/O(2) mixtures. We observed that the hydroperoxyl (HO(2)) radical plays a key role in the reaction kinetics at our input conditions (T ≥ 3000 K, P > 400 atm). The reaction mechanism observed is in good agreement with predictions of existing continuum-scale kinetic models for hydrogen combustion, and a transition of reaction mechanism is observed as we move from high pressure, low temperature to low pressure, high temperature. Since ReaxFF derives its parameters from quantum mechanical data and can simulate reaction pathways without any preconditioning, we believe that atomistic simulations through ReaxFF could be a useful tool in enhancing existing continuum-scale kinetic models for prediction of hydrogen combustion kinetics at high-pressure and high-temperature conditions, which otherwise is difficult to attain through experiments.