Evaluation of Sampling Algorithms Used for Bayesian Uncertainty Quantification of Molecular Dynamics Force Fields.

New Bayesian parameter estimation methods have the capability to enable more physically realistic and reliable molecular dynamics (MD) simulations by providing accurate estimates of uncertainties of force-field (FF) parameters and associated properties. However, the choice of which Bayesian parameter estimation algorithm to use has not been widely investigated, despite its impact on the effective exploration of parameter space. Here, using a case example of the Embedded Atom Method (EAM) FF parameters, we investigated the ramifications of several of the algorithm choices. We found that Ensemble Slice Sampling (ESS) and Affine-Invariant Ensemble Sampling (AIES) demonstrate a new level of superior performance, culminating in more accurate parameter and property estimations with tighter uncertainty bounds, compared to traditional methods such as Metropolis-Hastings (MH), Gradient Search (GS), and Uniform Random Sampler (URS). We demonstrate that Bayesian Uncertainty Quantification with ESS and AIES leads to significantly more accurate and reliable predictions of the FF parameters and properties. The results suggest that ESS and AIES should be used to obtain more accurate parameter and uncertainty estimations while providing deeper physical insights.

Probing the role of protein conformational changes in the mechanism of prenylated-FMN-dependent phenazine-1-carboxylic acid decarboxylase.

Phenazine-1-carboxylic acid decarboxylase (PhdA) is a prenylated-FMN-dependent (prFMN) enzyme belonging to the UbiD family of decarboxylases. Many UbiD-like enzymes catalyze (de)carboxylation reactions on aromatic rings and conjugated double bonds and are potentially valuable industrial catalysts. We have investigated the mechanism of PhdA using a slow turnover substrate, 2,3-dimethylquinoxaline-5-carboxylic acid (DQCA). Detailed analysis of the pH dependence and solvent deuterium isotope effects associated with the reaction uncovered unusual kinetic behavior. At low substrate concentrations, a substantial inverse solvent isotope effect (SIE) is observed on Vmax/KM of ∼ 0.5 when reaction rates of DQCA in H2O and D2O are compared. Under the same conditions, a normal SIE of 4.15 is measured by internal competition for proton transfer to the product. These apparently contradictory results indicate that the SIE values report on different steps in the mechanism. A proton inventory analysis of the reaction under Vmax/KM and Vmax conditions points to a "medium effect" as the source of the inverse SIE. Molecular dynamics simulations of the effect of D2O on PhdA structure support that D2O reduces the conformational lability of the enzyme and results in a more compact structure, akin to the active, "closed" conformer observed in crystal structures of some UbiD-like enzymes. Consistent with the simulations, PhdA was found to be more stable in D2O and to bind DQCA more tightly, leading to the observed rate enhancement under Vmax/KM conditions.

Data Driven Computational Design and Experimental Validation of Drugs for Accelerated Mitigation of Pandemic-like Scenarios.

Emerging pathogens are a historic threat to public health and economic stability. Current trial-and-error approaches to identify new therapeutics are often ineffective due to their inefficient exploration of the enormous small molecule design space. Here, we present a data-driven computational framework composed of hybrid evolutionary algorithms for evolving functional groups on existing drugs to improve their binding affinity toward the main protease (Mpro) of SARS-CoV-2. We show that combinations of functional groups and sites are critical to design drugs with improved binding affinity, which can be easily achieved using our framework by exploring a fraction of the available search space. Atomistic simulations and experimental validation elucidate that enhanced and prolonged interactions between functionalized drugs and Mpro residues result in their improved therapeutic value over that of the parental compound. Overall, this novel framework is extremely flexible and has the potential to rapidly design inhibitors for any protein with available crystal structures.

Data Driven Discovery of MOFs for Hydrogen Gas Adsorption.

Hydrogen gas (H2) is a clean and renewable energy source, but the lack of efficient and cost-effective storage materials is a challenge to its widespread use. Metal-organic frameworks (MOFs), a class of porous materials, have been extensively studied for H2 storage due to their tunable structural and chemical features. However, the large design space offered by MOFs makes it challenging to select or design appropriate MOFs with a high H2 storage capacity. To overcome these challenges, we present a data-driven computational approach that systematically designs new functionalized MOFs for H2 storage. In particular, we showcase the framework of a hybrid particle swarm optimization integrated genetic algorithm, grand canonical Monte Carlo (GCMC) simulations, and our in-house MOF structure generation code to design new MOFs with excellent H2 uptake. This automated, data driven framework adds appropriate functional groups to IRMOF-10 to improve its H2 adsorption capacity. A detailed analysis of the top selected MOFs, their adsorption isotherms, and MOF design rules to enhance H2 adsorption are presented. We found a functionalized IRMOF-10 with an enhanced H2 adsorption increased by ∼6 times compared to that of pure IRMOF-10 at 1 bar and 77 K. Furthermore, this study also utilizes machine learning and deep learning techniques to analyze a large data set of MOF structures and properties, in order to identify the key factors that influence hydrogen adsorption. The proof-of-concept that uses a machine learning/deep learning approach to predict hydrogen adsorption based on the identified structural and chemical properties of the MOF is demonstrated.

Development, Validation, and Applications of Nonbonded Interaction Parameters between Coarse-Grained Amino Acid and Water Models.

Interactions between amino acids and water play an important role in determining the stability and folding/unfolding, in aqueous solution, of many biological macromolecules, which affects their function. Thus, understanding the molecular-level interactions between water and amino acids is crucial to tune their function in aqueous solutions. Herein, we have developed nonbonded interaction parameters between the coarse-grained (CG) models of 20 amino acids and the one-site CG water model. The nonbonded parameters, represented using the 12-6 Lennard Jones (LJ) potential form, have been optimized using an artificial neural network (ANN)-assisted particle swarm optimization (PSO) (ANN-assisted PSO) method. All-atom (AA) molecular dynamics (MD) simulations of dipeptides in TIP3P water molecules were performed to calculate the Gibbs hydration free energies. The nonbonded force-field (FF) parameters between CG amino acids and the one-site CG water model were developed to accurately reproduce these energies. Furthermore, to test the transferability of these newly developed parameters, we calculated the hydration free energies of the analogues of the amino acid side chains, which showed good agreement with reported experimental data. Additionally, we show the applicability of these models by performing self-assembly simulations of peptide amphiphiles. Overall, these models are transferable and can be used to study the self-assembly of various biomaterials and biomolecules to develop a mechanistic understanding of these processes.

Supramolecular Peptide Nanostructures Regulate Catalytic Efficiency and Selectivity.

We report three constitutionally isomeric tetrapeptides, each comprising one glutamic acid (E) residue, one histidine (H) residue, and two lysine (KS ) residues functionalized with side-chain hydrophobic S-aroylthiooxime (SATO) groups. Depending on the order of amino acids, these amphiphilic peptides self-assembled in aqueous solution into different nanostructures:nanoribbons, a mixture of nanotoroids and nanoribbons, or nanocoils. Each nanostructure catalyzed hydrolysis of a model substrate, with the nanocoils exhibiting the greatest rate enhancement and the highest enzymatic efficiency. Coarse-grained molecular dynamics simulations, analyzed with unsupervised machine learning, revealed clusters of H residues in hydrophobic pockets along the outer edge of the nanocoils, providing insight for the observed catalytic rate enhancement. Finally, all three supramolecular nanostructures catalyzed hydrolysis of the l-substrate only when a pair of enantiomeric Boc-l/d-Phe-ONp substrates were tested. This study highlights how subtle molecular-level changes can influence supramolecular nanostructures, and ultimately affect catalytic efficiency.

A review of recent advances and applications of machine learning in tribology.

In tribology, a considerable number of computational and experimental approaches to understand the interfacial characteristics of material surfaces in motion and tribological behaviors of materials have been considered to date. Despite being useful in providing important insights on the tribological properties of a system, at different length scales, a vast amount of data generated from these state-of-the-art techniques remains underutilized due to lack of analysis methods or limitations of existing analysis techniques. In principle, this data can be used to address intractable tribological problems including structure-property relationships in tribological systems and efficient lubricant design in a cost and time effective manner with the aid of machine learning. Specifically, data-driven machine learning methods have shown potential in unraveling complicated processes through the development of structure-property/functionality relationships based on the collected data. For example, neural networks are incredibly effective in modeling non-linear correlations and identifying primary hidden patterns associated with these phenomena. Here we present several exemplary studies that have demonstrated the proficiency of machine learning in understanding these critical factors. A successful implementation of neural networks, supervised, and stochastic learning approaches in identifying structure-property relationships have shed light on how machine learning may be used in certain tribological applications. Moreover, ranging from the design of lubricants, composites, and experimental processes to studying fretting wear and frictional mechanism, machine learning has been embraced either independently or integrated with optimization algorithms by scientists to study tribology. Accordingly, this review aims at providing a perspective on the recent advances in the applications of machine learning in tribology. The review on referenced simulation approaches and subsequent applications of machine learning in experimental and computational tribology shall motivate researchers to introduce the revolutionary approach of machine learning in efficiently studying tribology.

Modelling drug adsorption in metal-organic frameworks: the role of solvent.

Solvent plays a key role in biological functions, catalysis, and drug delivery. Metal-organic frameworks (MOFs) due to their tunable functionalities, porosities and surface areas have been recently used as drug delivery vehicles. To investigate the effect of solvent on drug adsorption in MOFs, we have performed integrated computational and experimental studies in selected biocompatible MOFs, specifically, UiO-AZB, HKUST-1 (or CuBTC) and NH2-MIL-53(Al). The adsorption of three drugs, namely, 5-fluorouracil (5-FU), ibuprofen (IBU), and hydroxyurea (HU) were performed in the presence and absence of the ethanol. Our computational predictions, at 1 atmospheric pressure, showed a reasonable agreement with experimental studies performed in the presence of ethanol. We find that in the presence of ethanol the drug molecules were adsorbed at the interface of solvent and MOFs. Moreover, the computationally calculated adsorption isotherms suggested that the drug adsorption was driven by electrostatic interactions at lower pressures (<10-4 Pa). Our computational predictions in the absence of ethanol were higher compared to those in the presence of ethanol. The MOF-adsorbate interaction (U HA) energy decreased with decrease in the size of a drug molecule in all three MOFs at all simulated pressures. At high pressure the interaction energy increases with increase in the MOFs pore size as the number of molecules adsorbed increases. Thus, our research shows the important role played by solvent in drug adsorption and suggests that it is critical to consider solvent while performing computational studies.

A combined experimental and computational approach reveals how aromatic peptide amphiphiles self-assemble to form ion-conducting nanohelices.

Reported here is a combined experimental-computational strategy to determine structure-property-function relationships in persistent nanohelices formed by a set of aromatic peptide amphiphile (APA) tetramers with the general structure K S XEK S , where KS= S-aroylthiooxime modified lysine, X = glutamic acid or citrulline, and E = glutamic acid. In low phosphate buffer concentrations, the APAs self-assembled into flat nanoribbons, but in high phosphate buffer concentrations they formed nanohelices with regular twisting pitches ranging from 9-31 nm. Coarse-grained molecular dynamics simulations mimicking low and high salt concentrations matched experimental observations, and analysis of simulations revealed that increasing strength of hydrophobic interactions under high salt conditions compared with low salt conditions drove intramolecular collapse of the APAs, leading to nanohelix formation. Analysis of the radial distribution functions in the final self-assembled structures led to several insights. For example, comparing distances between water beads and beads representing hydrolysable KS units in the APAs indicated that the KS units in the nanohelices should undergo hydrolysis faster than those in the nanoribbons; experimental results verified this hypothesis. Simulation results also suggested that these nanohelices might display high ionic conductivity due to closer packing of carboxylate beads in the nanohelices than in the nanoribbons. Experimental results showed no conductivity increase over baseline buffer values for unassembled APAs, a slight increase (0.4 × 102 µS/cm) for self-assembled APAs under low salt conditions in their nanoribbon form, and a dramatic increase (8.6 × 102 µS/cm) under high salt conditions in their nanohelix form. Remarkably, under the same salt conditions, these self-assembled nanohelices conducted ions 5-10-fold more efficiently than several charged polymers, including alginate and DNA. These results highlight how experiments and simulations can be combined to provide insight into how molecular design affects self-assembly pathways; additionally, this work highlights how this approach can lead to discovery of unexpected properties of self-assembled nanostructures.

Machine learning approach for accurate backmapping of coarse-grained models to all-atom models.

Four different machine learning (ML) regression models: artificial neural network, k-nearest neighbors, Gaussian process regression and random forest were built to backmap coarse-grained models to all-atom models. The ML models showed better predictions than the existing backmapping approaches for selected structures, suggesting the applications of the ML models for backmapping.

Solvation dynamics of N-substituted acrylamide polymers and the importance for phase transition behavior.

Functional groups present in thermo-responsive polymers are known to play an important role in aqueous solutions by manifesting their coil-to-globule conformational transition in a specific temperature range. Understanding the role of these functional groups and their interactions with water is of great interest as it may allow us to control both the nature and temperature of this coil-to-globule transition. In this work, polyacrylamide (PAAm), poly(N-isopropylacrylamide) (PNIPAm), and poly(N-isopropylmethacrylamide) (PNIPMAm) solvated in water are studied with the goal of discovering the structure of the solvent and its interaction with these polymers in determining the polymer conformations. Specifically, all-atom molecular dynamics (MD) simulations were performed on polymer chains with 30 monomer units (30-mers) at 295 K, 310 K and 320 K, which is below and above the lower critical solution temperature (LCST) of PNIPAm (LCST = 305 K) and PNIPMAm (LCST = 315 K), respectively. The MD simulation trajectories suggest that changes in the functional groups in the backbone and side-chains alter the water solvation shell around the polymer. This results in a change in the residence time probability and hydrogen bond characteristics of water at simulated temperatures. Specifically, water molecules reside for longer times near PAAm (no LCST) and PNIPMAm (LCST = 315 K) chains as compared to PNIPAm. This might be one of the possible causes for the higher LCST of PNIPMAm as compared to that of PNIPAm. These results can guide experimentalists and theoreticians to design new polymer structures with tailor-made LCST transitions while controlling the water solvation shell around the functional group.

Machine-Learning Based Stacked Ensemble Model for Accurate Analysis of Molecular Dynamics Simulations.

Accurate, faster, and on-the-fly analysis of the molecular dynamics (MD) simulations trajectory becomes very critical during the discovery of new materials or while developing force-field parameters due to automated nature of these processes. Here to overcome the drawbacks of algorithm based analysis approaches, we have developed and utilized an approach that integrates machine-learning (ML) based stacked ensemble model (SEM) with MD simulations, for the first time. As a proof-of-concept, two SEMs were developed to analyze two dynamical properties of a water droplet, its contact angle, and hydrogen bonds. The two SEMs consisted of two layered networks of random forest, artificial neural network, support vector regression, Kernel ridge regression, and k-nearest neighbors ML models. The root-mean-square error values, uncertainty quantification, and sensitivity analysis of both the SEMs suggested that the final result was more accurate as compared to that of the individual ML models. This new computational framework is very general, robust, and has a huge potential in analyzing large size MD simulation trajectories as it can capture critical information very accurately.

Development of Transferable Nonbonded Interactions between Coarse-Grained Hydrocarbon and Water Models.

Interactions between water and hydrocarbons play a significant role in chemical, physical, and biological processes. Here, we present a set of force-field (FF) parameters that define the interactions between coarse-grained (CG) hydrocarbon models ( An , Y. J. Phys. Chem. B , 2018 , 122 , 7143 - 7153 ) and one-site water model ( Bejagam , K. K. J. Phys. Chem. B , 2018 , 122 , 1958 - 1971 ) developed in our recent work. The nonbonded FF interactions between various hydrocarbon beads and the water beads are represented by the 12-6 Lennard-Jones potential. The FF parameters were optimized to reproduce the experimentally measured Gibbs hydration free energies of selected hydrocarbon models (decane and hexadecane with 2:1 mapping scheme and nonane and pentadecane with 3:1 mapping scheme) and the interfacial tensions of decane and nonane models at 300 K. The predicted values of Gibbs hydration free energies of CG decane, hexadecane, nonane, and pentadecane models by the optimized FF parameters were within 8, 12, 11, and 4% of their corresponding experimental values, respectively. These new optimized FF parameters were transferable when used to calculate the Gibbs hydration free energies of different hydrocarbons ranging from pentane to heptadecane at 300 K (minimum error ∼0.5%, and maximum error ∼40.8%). Furthermore, the interfacial tensions of the CG hydrocarbon models calculated by using these new FF parameters showed good agreement with their corresponding experimental values at 300 K. Homogeneous mixtures of CG water and hydrocarbon models were able to exhibit the phase segregation during 1 µs. These new nonbonded interaction parameters were expected to be utilized in modeling the interactions between water and polymer backbones represented with hydrocarbon beads.

Machine-Learning Enabled New Insights into the Coil-to-Globule Transition of Thermosensitive Polymers Using a Coarse-Grained Model.

We present a computational framework that integrates coarse-grained (CG) molecular dynamics (MD) simulations and a data-driven machine-learning (ML) method to gain insights into the conformations of polymers in solutions. We employ this framework to study conformational transition of a model thermosensitive polymer, poly( N-isopropylacrylamide) (PNIPAM). Here, we have developed the first of its kind, a temperature-independent CG model of PNIPAM that can accurately predict its experimental lower critical solution temperature (LCST) while retaining the tacticity in the presence of an explicit water model. The CG model was extensively validated by performing CG MD simulations with different initial conformations, varying the radius of gyration of chain, the chain length, and the angle between the adjacent monomers of the initial configuration of PNIPAM (total simulation time = 90 µs). Moreover, for the first time, we utilize the nonmetric multidimensional scaling (NMDS) method, a data-driven ML approach, to gain further insights into the mechanisms and pathways of this coil-to-globule transition by analyzing CG MD simulation trajectories. NMDS analysis provides entirely new insights and shows multiple metastable states of PNIPAM during its coil-to-globule transition above the LCST.

Machine-Learned Coarse-Grained Models.

Optimizing force-field (FF) parameters to perform molecular dynamics (MD) simulations is a challenging and time-consuming process. We present a novel FF optimization framework that integrates MD simulations with particle swarm optimization (PSO) algorithm and artificial neural network (ANN). This new ANN-assisted PSO framework was used to develop transferable coarse-grained (CG) models for D2O and DMF as a proof of concept. The PSO algorithm was used to generate the set of input FF parameters for the MD simulations of the CG models of these solvents, which were optimized to reproduce their experimental properties. Herein, for the first time, a reverse approach was employed for on-the-fly training of the ANN model, where results (solvent properties) obtained from the MD simulations and their corresponding FF parameters were used as inputs and outputs, respectively. The ANN model was then required to predict a set of new FF parameters, which were tested for their ability to predict the desired experimental properties. This new framework can be extended to integrate any optimization algorithm with ANN and MD simulations to accelerate the FF development.

Development of New Transferable Coarse-Grained Models of Hydrocarbons.

We have utilized an approach that integrates molecular dynamics (MD) simulations with particle swarm optimization (PSO) to accelerate the development of coarse-grained (CG) models of hydrocarbons. Specifically, we have developed new transferable CG beads, which can be used to model the hydrocarbons (C5 to C17) and reproduce their experimental properties with good accuracy. First, the PSO method was used to develop the CG beads of the decane model represented with a 2:1 (2-2-2-2-2) mapping scheme. This was followed by the development of the nonane model described with hybrid 2-2-3-2 and 3:1 (3-3-3) mapping schemes. The force-field parameters for these three CG models were optimized to reproduce four experimentally observed properties including density, enthalpy of vaporization, surface tension, and self-diffusion coefficient at 300 K. The CG MD simulations conducted with these new CG models of decane and nonane, at different timesteps, for various system sizes, and at a range of different temperatures, were able to predict their density, enthalpy of vaporization, surface tension, self-diffusion coefficient, expansibility, and isothermal compressibility with good accuracy. Moreover, a comparison of structural features obtained from the CG MD simulations and the CG beads of mapped all-atom trajectories of decane and nonane showed very good agreement. To test the chemical transferability of these models, we have constructed the models for hydrocarbons ranging from pentane to heptadecane, by using different combinations of the CG beads of decane and nonane. The properties of pentane to heptadecane predicted by these new CG models showed excellent agreement with the experimental data.

PSO-Assisted Development of New Transferable Coarse-Grained Water Models.

We have employed two-to-one mapping scheme to develop three coarse-grained (CG) water models, namely, 1-, 2-, and 3-site CG models. Here, for the first time, particle swarm optimization (PSO) and gradient descent methods were coupled to optimize the force-field parameters of the CG models to reproduce the density, self-diffusion coefficient, and dielectric constant of real water at 300 K. The CG MD simulations of these new models conducted with various timesteps, for different system sizes, and at a range of different temperatures are able to predict the density, self-diffusion coefficient, dielectric constant, surface tension, heat of vaporization, hydration free energy, and isothermal compressibility of real water with excellent accuracy. The 1-site model is ∼3 and ∼4.5 times computationally more efficient than 2- and 3-site models, respectively. To utilize the speed of 1-site model and electrostatic interactions offered by 2- and 3-site models, CG MD simulations of 1:1 combination of 1- and 2-/3-site models were performed at 300 K. These mixture simulations could also predict the properties of real water with good accuracy. Two new CG models of benzene, consisting of beads with and without partial charges, were developed. All three water models showed good capacity to solvate these benzene models.

Development of non-bonded interaction parameters between graphene and water using particle swarm optimization.

New Lennard-Jones parameters have been developed to describe the interactions between atomistic model of graphene, represented by REBO potential, and five commonly used all-atom water models, namely SPC, SPC/E, SPC/Fw, SPC/Fd, and TIP3P/Fs by employing particle swarm optimization (PSO) method. These new parameters were optimized to reproduce the macroscopic contact angle of water on a graphene sheet. The calculated line tension was in the order of 10-11 J/m for the droplets of all water models. Our molecular dynamics simulations indicate the preferential orientation of water molecules near graphene-water interface with one OH bond pointing toward the graphene surface. Detailed analysis of simulation trajectories reveals the presence of water molecules with ≤∼1, ∼2, and ∼4 hydrogen bonds at the surface of air-water interface, graphene-water interface, and bulk region of the water droplet, respectively. Presence of water molecules with ≤∼1 and ∼2 hydrogen bonds suggest the existence of water clusters of different sizes at these interfaces. The trends observed in the libration, bending, and stretching bands of the vibrational spectra are closely associated with these structural features of water. The inhomogeneity in hydrogen bond network of water at the air-water and graphene-water interface is manifested by broadening of the peaks in the libration band for water present at these interfaces. The stretching band for the molecules in water droplet shows a blue shift as compared to the pure bulk water, which conjecture the presence of weaker hydrogen bond network in a droplet. © 2017 Wiley Periodicals, Inc.

Water ordering controls the dynamic equilibrium of micelle-fibre formation in self-assembly of peptide amphiphiles.

Understanding the role of water in governing the kinetics of the self-assembly processes of amphiphilic peptides remains elusive. Here, we use a multistage atomistic-coarse-grained approach, complemented by circular dichroism/infrared spectroscopy and dynamic light scattering experiments to highlight the dual nature of water in driving the self-assembly of peptide amphiphiles (PAs). We show computationally that water cage formation and breakage near the hydrophobic groups control the fusion dynamics and aggregation of PAs in the micellar stage. Simulations also suggest that enhanced structural ordering of vicinal water near the hydrophilic amino acids shifts the equilibrium towards the fibre phase and stimulates structure and order during the PA assembly into nanofibres. Experiments validate our simulation findings; the measured infrared O-H bond stretching frequency is reminiscent of an ice-like bond which suggests that the solvated water becomes increasingly ordered with time in the assembled peptide network, thus shedding light on the role of water in a self-assembly process.

Metal-induced rapid transformation of diamond into single and multilayer graphene on wafer scale.

The degradation of intrinsic properties of graphene during the transfer process constitutes a major challenge in graphene device fabrication, stimulating the need for direct growth of graphene on dielectric substrates. Previous attempts of metal-induced transformation of diamond and silicon carbide into graphene suffers from metal contamination and inability to scale graphene growth over large area. Here, we introduce a direct approach to transform polycrystalline diamond into high-quality graphene layers on wafer scale (4 inch in diameter) using a rapid thermal annealing process facilitated by a nickel, Ni thin film catalyst on top. We show that the process can be tuned to grow single or multilayer graphene with good electronic properties. Molecular dynamics simulations elucidate the mechanism of graphene growth on polycrystalline diamond. In addition, we demonstrate the lateral growth of free-standing graphene over micron-sized pre-fabricated holes, opening exciting opportunities for future graphene/diamond-based electronics.