Collaborative Assessment of Molecular Geometries and Energies from the Open Force Field.

Force fields form the basis for classical molecular simulations, and their accuracy is crucial for the quality of, for instance, protein-ligand binding simulations in drug discovery. The huge diversity of small-molecule chemistry makes it a challenge to build and parameterize a suitable force field. The Open Force Field Initiative is a combined industry and academic consortium developing a state-of-the-art small-molecule force field. In this report, industry members of the consortium worked together to objectively evaluate the performance of the force fields (referred to here as OpenFF) produced by the initiative on a combined public and proprietary dataset of 19,653 relevant molecules selected from their internal research and compound collections. This evaluation was important because it was completely blind; at most partners, none of the molecules or data were used in force field development or testing prior to this work. We compare the Open Force Field "Sage" version 2.0.0 and "Parsley" version 1.3.0 with GAFF-2.11-AM1BCC, OPLS4, and SMIRNOFF99Frosst. We analyzed force-field-optimized geometries and conformer energies compared to reference quantum mechanical data. We show that OPLS4 performs best, and the latest Open Force Field release shows a clear improvement compared to its predecessors. The performance of established force fields such as GAFF-2.11 was generally worse. While OpenFF researchers were involved in building the benchmarking infrastructure used in this work, benchmarking was done entirely in-house within industrial organizations and the resulting assessment is reported here. This work assesses the force field performance using separate benchmarking steps, external datasets, and involving external research groups. This effort may also be unique in terms of the number of different industrial partners involved, with 10 different companies participating in the benchmark efforts.

Implementing reactivity in molecular dynamics simulations with harmonic force fields.

The simulation of chemical reactions and mechanical properties including failure from atoms to the micrometer scale remains a longstanding challenge in chemistry and materials science. Bottlenecks include computational feasibility, reliability, and cost. We introduce a method for reactive molecular dynamics simulations using a clean replacement of non-reactive classical harmonic bond potentials with reactive, energy-conserving Morse potentials, called the Reactive INTERFACE Force Field (IFF-R). IFF-R is compatible with force fields for organic and inorganic compounds such as IFF, CHARMM, PCFF, OPLS-AA, and AMBER. Bond dissociation is enabled by three interpretable Morse parameters per bond type and zero energy upon disconnect. Use cases for bond breaking in molecules, failure of polymers, carbon nanostructures, proteins, composite materials, and metals are shown. The simulation of bond forming reactions is included via template-based methods. IFF-R maintains the accuracy of the corresponding non-reactive force fields and is about 30 times faster than prior reactive simulation methods.

Accurate Force Fields for Atomistic Simulations of Oxides, Hydroxides, and Organic Hybrid Materials up to the Micrometer Scale.

The simulation of metals, oxides, and hydroxides can accelerate the design of therapeutics, alloys, catalysts, cement-based materials, ceramics, bioinspired composites, and glasses. Here we introduce the INTERFACE force field (IFF) and surface models for α-Al2O3, α-Cr2O3, α-Fe2O3, NiO, CaO, MgO, ß-Ca(OH)2, ß-Mg(OH)2, and ß-Ni(OH)2. The force field parameters are nonbonded, including atomic charges for Coulomb interactions, Lennard-Jones (LJ) potentials for van der Waals interactions with 12-6 and 9-6 options, and harmonic bond stretching for hydroxide ions. The models outperform DFT calculations and earlier atomistic models (Pedone, ReaxFF, UFF, CLAYFF) up to 2 orders of magnitude in reliability, compatibility, and interpretability due to a quantitative representation of chemical bonding consistent with other compounds across the periodic table and curated experimental data for validation. The IFF models exhibit average deviations of 0.2% in lattice parameters, <10% in surface energies (to the extent known), and 6% in bulk moduli relative to experiments. The parameters and models can be used with existing parameters for solvents, inorganic compounds, organic compounds, biomolecules, and polymers in IFF, CHARMM, CVFF, AMBER, OPLS-AA, PCFF, and COMPASS, to simulate bulk oxides, hydroxides, electrolyte interfaces, and multiphase, biological, and organic hybrid materials at length scales from atoms to micrometers. The nonbonded character of the models also enables the analysis of mixed oxides, glasses, and certain chemical reactions, and well-performing nonbonded models for silica phases, SiO2, are introduced. Automated model building is available in the CHARMM-GUI Nanomaterial Modeler. We illustrate applications of the models to predict the structure of mixed oxides, and energy barriers of ion migration, as well as binding energies of water and organic molecules in outstanding agreement with experimental data and calculations at the CCSD(T) level. Examples of model building for hydrated, pH-sensitive oxide surfaces to simulate solid-electrolyte interfaces are discussed.

Effective potentials between nanoparticles in suspension.

Results of molecular dynamics simulations are presented for the pair distribution function between nanoparticles in an explicit solvent as a function of nanoparticle diameter and interaction strength between the nanoparticle and solvent. The effect of including the solvent explicitly is demonstrated by comparing the pair distribution function of nanoparticles to that in an implicit solvent. The nanoparticles are modeled as a uniform distribution of Lennard-Jones particles, while the solvent is represented by standard Lennard-Jones particles. The diameter of the nanoparticle is varied from 10 to 25 times that of the solvent for a range of nanoparticle volume fractions. As the strength of the interactions between nanoparticles and the solvent increases, the solvent layer surrounding the nanoparticle is formed which increases the effective radii of the nanoparticles. The pair distribution functions are inverted using the Ornstein-Zernike integral equation to determine an effective pair potential between the nanoparticles mediated by the introduction of an explicit solvent.

Closed-Form Coexistence Equation for Phase Separation of Polymeric Mixtures in Dissipative Particle Dynamics.

To date, no extensive study of the phase diagram for binary fluid mixtures in dissipative particle dynamics (DPD) has been published. This is especially pertinent for newer parameterization schemes where the self-self interaction, or the effective volume, of different particle types is varied. This work presents an exhaustive study of the parameter space concerning DPD particles with soft interaction potentials. Moreover, we propose a closed-form coexistence equation or binodal curve that is inspired by the Flory-Huggins model. This equation describes the phase diagram of all binary mixtures made up out of monomers, homopolymers, and the mixtures thereof when self-self interactions are varied. The mean absolute percentage error (MAPE) of the equation on simulated data, including validation simulations, is 1.02%. The equation can a priori predict the phase separation of mixtures using only DPD interaction parameters. The proposed coexistence equation can therefore be used to directly validate interaction parameters resulting from novel parameterization schemes, including coarse graining and equations of state, without the need for additional simulations. Finally, it is shown that the choice of bond potential markedly influences phase behavior.

Mesoscale hydrodynamics via stochastic rotation dynamics: comparison with Lennard-Jones fluid.

Stochastic rotation dynamics (SRD) is a relatively recent technique, closely related to lattice Boltzmann, for capturing hydrodynamic fluid flow at the mesoscale. The SRD method is based on simple constituent fluid particle interactions and dynamics. Here we parametrize the SRD fluid to provide a one to one match in the shear viscosity of a Lennard-Jones fluid and present viscosity measurements for a range of such parameters. We demonstrate how to apply the Müller-Plathe reverse perturbation method for determining the shear viscosity of the SRD fluid and discuss how finite system size and momentum exchange rates effect the measured viscosity. The implementation and performance of SRD in a parallel molecular dynamics code is also described.

Shear thinning of nanoparticle suspensions.

Results of large scale nonequilibrium molecular dynamics simulations are presented for nanoparticles in an explicit solvent. The nanoparticles are modeled as a uniform distribution of Lennard-Jones particles, while the solvent is represented by standard Lennard-Jones particles. We present results for the shear rheology of spherical nanoparticles of diameter 10 times that of the solvent for a range of nanoparticle volume fractions. By varying the strength of the interactions between nanoparticles and with the solvent, this system can be used to model colloidal gels and glasses as well as hard spherelike nanoparticles. Effect of including the solvent explictly is demonstrated by comparing the pair correlation function of nanoparticles to that in an implicit solvent. The shear rheology for dumbbell nanoparticles made of two fused spheres is similar to that of single nanoparticle.

Simulation of the mechanical strength of a single collagen molecule.

We perform atomistic simulations on a single collagen molecule to determine its intrinsic molecular strength. A tensile pull simulation to determine the tensile strength and Young's modulus is performed, and a simulation that separates two of the three helices of collagen examines the internal strength of the molecule. The magnitude of the calculated tensile forces is consistent with the strong forces of bond stretching and angle bending that are involved in the tensile deformation. The triple helix unwinds with increasing tensile force. Pulling apart the triple helix has a smaller, oscillatory force. The oscillations are due to the sequential separation of the hydrogen-bonded helices. The force rises due to reorienting the residues in the direction of the separation force. The force drop occurs once the hydrogen bond between residues on different helices break and the residues separate.

Liquid-vapor coexistence for nanoparticles of various size.

We present molecular dynamics simulations of the liquid-vapor phase coexistence of pure nanoparticle systems with three different model nanoparticle interactions. Our simulations show that the form of the interaction potential between nanoparticles strongly influences their coexistence behavior. For nanoparticles interacting with an integrated Lennard-Jones potential, the critical temperature and critical density increase with increasing particle size. In contrast, nanoparticles interacting via a Lennard-Jones potential shifted to the surface of the nanoparticle do not exhibit the expected size dependence of the phase diagram. For this model, the critical temperature decreases with increasing nanoparticle size. Similar results were observed for composite nanoparticles, with the interactions truncated at a finite distance.

A nanobody-based tracer targeting DPP6 for non-invasive imaging of human pancreatic endocrine cells.

There are presently no reliable ways to quantify endocrine cell mass (ECM) in vivo, which prevents an accurate understanding of the progressive beta cell loss in diabetes or following islet transplantation. To address this unmet need, we coupled RNA sequencing of human pancreatic islets to a systems biology approach to identify new biomarkers of the endocrine pancreas. Dipeptidyl-Peptidase 6 (DPP6) was identified as a target whose mRNA expression is at least 25-fold higher in human pancreatic islets as compared to surrounding tissues and is not changed by proinflammatory cytokines. At the protein level, DPP6 localizes only in beta and alpha cells within the pancreas. We next generated a high-affinity camelid single-domain antibody (nanobody) targeting human DPP6. The nanobody was radiolabelled and in vivo SPECT/CT imaging and biodistribution studies were performed in immunodeficient mice that were either transplanted with DPP6-expressing Kelly neuroblastoma cells or insulin-producing human EndoC-ßH1 cells. The human DPP6-expressing cells were clearly visualized in both models. In conclusion, we have identified a novel beta and alpha cell biomarker and developed a tracer for in vivo imaging of human insulin secreting cells. This provides a useful tool to non-invasively follow up intramuscularly implanted insulin secreting cells.

Multiscale Materials Modeling in an Industrial Environment.

In this review, we sketch the materials modeling process in industry. We show that predictive and fast modeling is a prerequisite for successful participation in research and development processes in the chemical industry. Stable and highly automated workflows suitable for handling complex systems are a must. In particular, we review approaches to build and parameterize soft matter systems. By satisfying these prerequisites, efficiency for the development of new materials can be significantly improved, as exemplified here for formulation polymer development. This is in fact in line with recent Materials Genome Initiative efforts sponsored by the US government. Valuable contributions to product development are possible today by combining existing modeling techniques in an intelligent fashion, provided modeling and experiment work hand in hand.

Application of Ewald summations to long-range dispersion forces.

Results illustrating the effects of using explicit summation terms for the r(-6) dispersion term on the interfacial properties of a Lennard-Jones fluid and SPC/E water are presented. For the Lennard-Jones fluid, we find that the use of long-range summations, even with a short "crossover radius," yields results that are consistent with simulations using large cutoff radii. Simulations of SPC/E water demonstrate that the long-range dispersion forces are of secondary importance to the Coulombic forces. In both cases, we find that the ratio of the box size L( parallel) to the crossover radius r(c) (k) plays an important role in determining the magnitude of the long-range dispersion correction, although its effect is secondary when Coulombic interactions are also present.