Force Field Benchmark of Organic Liquids. 2. Gibbs Energy of Solvation

Comparative assessment of physics-based in silico methods to calculate relative solubilities

Relative solubilities, i.e. whether a given molecule is more soluble in one solvent compared to others, is a critical parameter for pharmaceutical and agricultural formulation development and chemical synthesis, material science, and environmental chemistry. In silico predictions of this crucial variable can help reducing experiments, waste of solvents and synthesis optimization. In this study, we evaluate the performance of different physics-based methods for predicting relative solubilities. Our assessment involves quantum mechanics-based COSMO-RS and molecular dynamics-based free energy methods using OPLS4, the open-source OpenFF Sage, and GAFF force fields, spanning over 200 solvent-solute combinations. Our investigation highlights the important role of compound multimerization, an effect which must be accounted for to obtain accurate relative solubility predictions. The performance landscape of these methods is varied, with significant differences in precision depending on both the method used and the solute considered, thereby offering an improved understanding of the predictive power of physics-based methods in chemical research.

Martini on the Rocks: Can a Coarse-Grained Force Field Model Crystals?

Computational chemistry is an important tool in numerous scientific disciplines, including drug discovery and structural biology. Coarse-grained models offer simple representations of molecular systems that enable simulations of large-scale systems. Because there has been an increase in the adoption of such models for simulations of biomolecular systems, critical evaluation is warranted. Here, the stability of the amyloid peptide and organic crystals is evaluated using the Martini 3 coarse-grained force field. The crystals change shape drastically during the simulations. Radial distribution functions show that the distance between backbone beads in β-sheets increases by ∼1 Å, breaking the crystals. The melting points of organic compounds are much too low in the Martini force field. This suggests that Martini 3 lacks the specific interactions needed to accurately simulate peptides or organic crystals without imposing artificial restraints. The problems may be exacerbated by the use of the 12-6 potential, suggesting that a softer potential could improve this model for crystal simulations.

Designing solvent systems using self-evolving solubility databases and graph neural networks

Designing solvent systems is key to achieving the facile synthesis and separation of desired products from chemical processes, so many machine learning models have been developed to predict solubilities. However, breakthroughs are needed to address deficiencies in the model's predictive accuracy and generalizability; this can be addressed by expanding and integrating experimental and computational solubility databases. To maximize predictive accuracy, these two databases should not be trained separately, and they should not be simply combined without reconciling the discrepancies from different magnitudes of errors and uncertainties. Here, we introduce self-evolving solubility databases and graph neural networks developed through semi-supervised self-training approaches. Solubilities from quantum-mechanical calculations are referred to during semi-supervised learning, but they are not directly added to the experimental database. Dataset augmentation is performed from 11 637 experimental solubilities to >900 000 data points in the integrated database, while correcting for the discrepancies between experiment and computation. Our model was successfully applied to study solvent selection in organic reactions and separation processes. The accuracy (mean absolute error around 0.2 kcal mol-1 for the test set) is quantitatively useful in exploring Linear Free Energy Relationships between reaction rates and solvation free energies for 11 organic reactions. Our model also accurately predicted the partition coefficients of lignin-derived monomers and drug-like molecules. While there is room for expanding solubility predictions to transition states, radicals, charged species, and organometallic complexes, this approach will be attractive to predictive chemistry areas where experimental, computational, and other heterogeneous data should be combined.

Molecular dynamics study on micelle-small molecule interactions: developing a strategy for an extensive comparison

Theoretical predictions of the solubilizing capacity of micelles and vesicles present in intestinal fluid are important for the development of new delivery techniques and bioavailability improvement. A balance between accuracy and computational cost is a key factor for an extensive study of numerous compounds in diverse environments. In this study, we aimed to determine an optimal molecular dynamics (MD) protocol to evaluate small-molecule interactions with micelles composed of bile salts and phospholipids. MD simulations were used to produce free energy profiles for three drug molecules (danazol, probucol, and prednisolone) and one surfactant molecule (sodium caprate) as a function of the distance from the colloid center of mass. To address the challenges associated with such tasks, we compared different simulation setups, including freely assembled colloids versus pre-organized spherical micelles, full free energy profiles versus only a few points of interest, and a coarse-grained model versus an all-atom model. Our findings demonstrate that combining these techniques is advantageous for achieving optimal performance and accuracy when evaluating the solubilization capacity of micelles. All-atom (AA) and coarse-grained (CG) umbrella sampling (US) simulations and point-wise free energy (FE) calculations were compared to their efficiency to computationally analyze the solubilization of active pharmaceutical ingredients in intestinal fluid colloids.

Predicting absolute aqueous solubility by applying a machine learning model for an artificially liquid-state as proxy for the solid-state

In this study, we use machine learning algorithms with QM-derived COSMO-RS descriptors, along with Morgan fingerprints, to predict the absolute solubility of drug-like compounds. The QM-derived descriptors account for the molecular properties of the solute, i.e., the solute-solute interactions in an artificial-liquid-state (super-cooled liquid), and the solute-solvent interactions in solution. We employ two main approaches to predict solubility: (i) a hypothetical pathway that involves melting the solute at room temperature T = T¯ ([Formula: see text]) and mixing the artificially liquid solute into the solvent ([Formula: see text]). In this approach [Formula: see text] is predicted using machine learning models, and the [Formula: see text] is obtained from COSMO-RS calculations; (ii) direct solubility prediction using machine learning algorithms. The models were trained on a large number of Bayer in-house compounds for which water solubility data is available at physiological pH of 6.5 and ambient temperature. We also evaluated our models using external datasets from a solubility challenge. Our models present great improvements compared to the absolute solubility prediction with the QSAR model for the artificial liquid state as implemented in the COSMOtherm software, for both in-house and external datasets. We are furthermore able to demonstrate the superiority of QM-derived descriptors compared to cheminformatics descriptors. We finally present low-cost alternative models using fragment-based COSMOquick calculations with only marginal reduction in the quality of predicted solubility.

Simulations of Amyloid-Forming Peptides in the Crystal State

There still is little treatment available for amyloid diseases, despite their significant impact on individuals and the social and economic implications for society. One reason for this is that the physical nature of amyloid formation is not understood sufficiently well. Therefore, fundamental research at the molecular level remains necessary to support the development of therapeutics. A few structures of short peptides from amyloid-forming proteins have been determined. These can in principle be used as scaffolds for designing aggregation inhibitors. Attempts to this end have often used the tools of computational chemistry, in particular molecular simulation. However, few simulation studies of these peptides in the crystal state have been presented so far. Hence, to validate the capability of common force fields (AMBER19SB, CHARMM36m, and OPLS-AA/M) to yield insight into the dynamics and structural stability of amyloid peptide aggregates, we have performed molecular dynamics simulations of twelve different peptide crystals at two different temperatures. From the simulations, we evaluate the hydrogen bonding patterns, the isotropic B-factors, the change in energy, the Ramachandran plots, and the unit cell parameters and compare the results with the crystal structures. Most crystals are stable in the simulations but for all force fields there is at least one that deviates from the experimental crystal, suggesting more work is needed on these models.

Role of Crystal Disorder and Mechanoactivation in Solid-State Stability of Pharmaceuticals

Common energy-intensive processes applied in oral solid dosage development, such as milling, sieving, blending, compaction, etc. generate particles with surface and bulk crystal disorder. An intriguing aspect of the generated crystal disorder is its evolution and repercussion on the physical- and chemical stabilities of drugs. In this review, we firstly examine the existing literature on crystal disorder and its implications on solid-state stability of pharmaceuticals. Secondly, we discuss the key aspects related to the generation and evolution of crystal disorder, dynamics of the disordered/amorphous phase, analytical techniques to measure/quantify them, and approaches to model the disordering propensity from first principles. The main objective of this compilation is to provide special impetus to predict or model the chemical degradation(s) resulting from processing-induced manifestation in bulk solid manufacturing. Finally, a generic workflow is proposed that can be useful to investigate the relevance of crystal disorder on the degradation of pharmaceuticals during stability studies. The present review will cater to the requirements for developing physically- and chemically stable drugs, thereby enabling early and rational decision-making during candidate screening and in assessing degradation risks associated with formulations and processing.

Development of a QM/MM(ABEEM) method combined with a polarizable force field to investigate the excision reaction mechanism of damaged thymine

This paper focuses on the development of a quantum mechanics/molecular mechanics method using the ABEEM polarizable force field (QM/MM(ABEEM) method) to investigate the excision reaction mechanism of damaged thymine. This method does not simply combine the QM method with the polarizable force field. A valence electronegativity piecewise function with the distance between atoms as a variable is introduced to describe the atomic partial charges, and changes greatly during the reaction process. At the same time, the charge transfer effect is treated using the condition of local charge conservation. Compared with the traditional QM/MM method, the QM/MM(ABEEM) method can more accurately simulate the polarization effect and charge transfer effect in the reaction process. Focusing on the controversial problems of the excision of damaged bases, six reaction pathways were designed for monofunctional and difunctional deglycosylation of neutral bases and protonated bases. The results show that the QM/MM(ABEEM) method accurately simulates the polarization effect, charge transfer effect, activation energy and other properties of the reaction process. The process in which the active residue Asp activates the nucleophile H₂O to attack the protonated base is the preferred path. The average activation energy and free activation energy of the protonated base are 7.00-14.00 kcal mol^-1 lower than that of the neutral base. The study in this paper is helpful to understand the mechanism of repair enzymes in repairing bases.

An Imbalance in the Force: The Need for Standardized Benchmarks for Molecular Simulation

Force fields (FFs) for molecular simulation have been under development for more than half a century. As with any predictive model, rigorous testing and comparisons of models critically depends on the availability of standardized data sets and benchmarks. While such benchmarks are rather common in the fields of quantum chemistry, this is not the case for empirical FFs. That is, few benchmarks are reused to evaluate FFs, and development teams rather use their own training and test sets. Here we present an overview of currently available tests and benchmarks for computational chemistry, focusing on organic compounds, including halogens and common ions, as FFs for these are the most common ones. We argue that many of the benchmark data sets from quantum chemistry can in fact be reused for evaluating FFs, but new gas phase data is still needed for compounds containing phosphorus and sulfur in different valence states. In addition, more nonequilibrium interaction energies and forces, as well as molecular properties such as electrostatic potentials around compounds, would be beneficial. For the condensed phases there is a large body of experimental data available, and tools to utilize these data in an automated fashion are under development. If FF developers, as well as researchers in artificial intelligence, would adopt a number of these data sets, it would become easier to compare the relative strengths and weaknesses of different models and to, eventually, restore the balance in the force.

Force fields optimized against experimental data for large compound families using CombiFF: Validation considering non-target properties and polyfunctional compounds

The CombiFF scheme is a workflow for the automated calibration of force-field parameters against condensed-phase experimental data considering simultaneously entire classes of organic molecules. The main steps of this scheme are: (i) selection of a molecule family; (ii) enumeration of all isomers; (iii) query for experimental data; (iv) automatic construction of the molecular topologies; (v) iterative refinement of the force-field parameters considering the entire family. In two recent articles, CombiFF was applied to the design of GROMOS-compatible united-atom force fields for the saturated acyclic haloalkanes and for saturated acyclic compounds involving eight common chemical functional groups of oxygen and nitrogen. This calibration and the subsequent initial validation involved two limitations: (i) the experimental data considered was restricted to values for the pure-liquid density ρ_liq and the vaporization enthalpy ΔH_vap of the compounds; (ii) beyond monofunctional compounds, the training set only involved homo-polyhaloalkanes (possibly mixing halogen types) in the first study, and homo-polyfunctional compounds of the considered oxygen or nitrogen functional groups (no mixing of different group types) in the second one. The goal of this article is to further test the accuracy of CombiFF-generated force fields by extending the validation to: (i) nine additional properties that were not used as optimization targets (pure-liquid thermodynamic, dielectric and transport properties, as well as solvation properties); (ii) hetero-polyfunctional molecules that were not included in the calibration and initial validation sets. The results for the nine additional properties show good agreement with experiment, except for the shear viscosity and the dielectric permittivity. There, larger discrepancies are observed, likely due to the united-atom representation adopted for the aliphatic groups and to the implicit treatment of electronic polarization effects. The results for the hetero-polyfunctional molecules also show reasonable agreement with experiment in terms of the monitored properties.

Explainable Solvation Free Energy Prediction Combining Graph Neural Networks with Chemical Intuition

The prediction of a molecule's solvation Gibbs free (ΔG_solv) energy in a given solvent is an important task which has traditionally been carried out via quantum chemical continuum methods or force field-based molecular simulations. Machine learning (ML) and graph neural networks in particular have emerged as powerful techniques for elucidating structure-property relationships. This work presents a graph neural network (GNN) for the prediction of ΔG_solv which, in addition to encoding typical atom and bond-level features, incorporates chemically intuitive, solvation-relevant parameters into the featurization process: semiempirical partial atomic charges and solvent dielectric constant. Solute-solvent interactions are included via an interaction map layer which can be visualized to examine solubility-enhancing or -decreasing interactions learnt by the model. On a test set of small organic molecules, our GNN predicts ΔG_solv in water and cyclohexane with an accuracy comparable to polarizable and ab initio generated force field methods [mean absolute error (MAE) = 0.4 and 0.2 kcal mol^-1, respectively], without the need for any molecular simulation. For the FreeSolv data set of hydration free energies, the test MAE is 0.7 kcal mol^-1. Interpretability and applicability of the model is highlighted through several examples including rationalizing the increased solubility of modified diaminoanthraquinones in organic solvents. The clear explanations afforded by our GNN allow for easy understanding of the model's predictions, giving the experimental chemist confidence in employing ML models toward more optimized synthetic routes.

Comparison of the United- and All-Atom Representations of (Halo)alkanes Based on Two Condensed-Phase Force Fields Optimized against the Same Experimental Data Set

The level of accuracy that can be achieved by a force field is influenced by choices made in the interaction-function representation and in the relevant simulation parameters. These choices, referred to here as functional-form variants (FFVs), include for example the model resolution, the charge-derivation procedure, the van der Waals combination rules, the cutoff distance, and the treatment of the long-range interactions. Ideally, assessing the effect of a given FFV on the intrinsic accuracy of the force-field representation requires that only the specific FFV is changed and that this change is performed at an optimal level of parametrization, a requirement that may prove extremely challenging to achieve in practice. Here, we present a first attempt at such a comparison for one specific FFV, namely the choice of a united-atom (UA) versus an all-atom (AA) resolution in a force field for saturated acyclic (halo)alkanes. Two force-field versions (UA vs AA) are optimized in an automated way using the CombiFF approach against 961 experimental values for the pure-liquid densities ρ_liq and vaporization enthalpies ΔH_vap of 591 compounds. For the AA force field, the torsional and third-neighbor Lennard-Jones parameters are also refined based on quantum-mechanical rotational-energy profiles. The comparison between the UA and AA resolutions is also extended to properties that have not been included as parameterization targets, namely the surface-tension coefficient γ, the isothermal compressibility κ_T, the isobaric thermal-expansion coefficient α_P, the isobaric heat capacity c_P, the static relative dielectric permittivity ϵ, the self-diffusion coefficient D, the shear viscosity η, the hydration free energy ΔG_wat, and the free energy of solvation ΔG_che in cyclohexane. For the target properties ρ_liq and ΔH_vap, the UA and AA resolutions reach very similar levels of accuracy after optimization. For the nine other properties, the AA representation leads to more accurate results in terms of η; comparably accurate results in terms of γ, κ_T, α_P, ϵ, D, and ΔG_che; and less accurate results in terms of c_P and ΔG_wat. This work also represents a first step toward the calibration of a GROMOS-compatible force field at the AA resolution.

Development of QM/MM (ABEEM polarizable force field) method to simulate the excision reaction mechanism of damaged cytosine

DNA damages are regarded as having harmful effects on cell. The base excision repair mechanism combats these effects by removing damaged bases. The deglycosylation mechanism of excising damaged bases by DNA glycosylase and the state of the leaving base have been controversial. The enzymatic reaction of DNA glycosylase to remove the damaged bases involves not only the formation and breaking of chemical bonds, but also complex polarization effect and charge transfer, which cannot be accurately simulated by the QM/MM method combined with the fixed charge force field. This work has developed the ABEEM fluctuating polarizable force field combining with the QM method, that is (QM/MM[ABEEM]), to accurately simulate the proton transfer, charge transfer and the charge distribution. The piecewise function is used as the valence-state electronegativity in the QM/MM (ABEEM) to realize the accurate fitting of the charge distribution in reaction. And the charge transfer is accurately simulated by the local charge conservation conditions. Four deglycosylation mechanisms including the monofunctional and difunctional mechanisms of four neutral and protonated cytosine derivatives are explored. It is confirmed that the monofunctional mechanism of Asp-activated nucleophile water is a better deglycosylation mechanism and the base is protonated before the reaction occurs. Protonization of the base reduced the activation energy by 10.00-17.00 kcal/mol. Asp provides the necessary charge for the reaction, and DNA glycosylase preferentially cleaves ɛC. This work provides a theoretical basis for the research of excising damaged bases by DNA glycosylase.

Automated Molecular Cluster Growing for Explicit Solvation by Efficient Force Field and Tight Binding Methods

An automated and broadly applicable workflow for the description of solvation effects in an explicit manner is introduced. This method, termed quantum cluster growth (QCG), is based on the semiempirical GFN2-xTB/GFN-FF methods, enabling efficient geometry optimizations and MD simulations. Fast structure generation is provided using the intermolecular force field xTB-IFF. Additionally, the approach uses an efficient implicit solvation model for the electrostatic embedding of the growing clusters. The novel QCG procedure presents a robust cluster generation tool for subsequent application of higher-level (e.g., DFT) methods to study solvation effects on molecular geometries explicitly or to average spectroscopic properties over cluster ensembles. Furthermore, the computation of the solvation free energy with a supermolecular approach can be carried out with QCG. The underlying growing process is physically motivated by computing the leading-order solute-solvent interactions first and can account for conformational and chemical changes due to solvation for low-energy barrier processes. The conformational space is explored with the NCI-MTD algorithm as implemented in the CREST program, using a combination of metadynamics and MD simulations. QCG with GFN2-xTB yields realistic solution geometries and reasonable solvation free energies for various systems without introducing many empirical parameters. Computed IR spectra of some solutes with QCG show a better match to the experimental data compared to well-established implicit solvation models.

MLSolvA: solvation free energy prediction from pairwise atomistic interactions by machine learning

Recent advances in machine learning technologies and their applications have led to the development of diverse structure-property relationship models for crucial chemical properties. The solvation free energy is one of them. Here, we introduce a novel ML-based solvation model, which calculates the solvation energy from pairwise atomistic interactions. The novelty of the proposed model consists of a simple architecture: two encoding functions extract atomic feature vectors from the given chemical structure, while the inner product between the two atomistic feature vectors calculates their interactions. The results of 6239 experimental measurements achieve outstanding performance and transferability for enlarging training data owing to its solvent-non-specific nature. An analysis of the interaction map shows that our model has significant potential for producing group contributions on the solvation energy, which indicates that the model provides not only predictions of target properties but also more detailed physicochemical insights.

Rational Design of Nonbonded Point Charge Models for Divalent Metal Cations with Lennard-Jones 12-6 Potential

Exploring a metal-involved biochemical process at a molecular level often requires a reliable description of metal properties in aqueous solution by classical nonbonded models. An additional C₄ term for considering ion-induced dipole interactions was previously proposed to supplement the widely used Lennard-Jones 12-6 potential (known as the 12-6-4 LJ-type model) with good accuracy. Here, we demonstrate an alternative to modeling divalent metal cations (M²⁺) with the traditional 12-6 LJ potential by developing nonbonded point charge models for use with 11 water models: TIP3P, SPC/E, SPC/E_b, TIP4P-Ew, TIP4P-D, and TIP4P/2005 and the more recent OPC3, TIP3P-FB, OPC, TIP4P-FB, and a99SB-disp. Our designed models simultaneously reproduce the experimental hydration free energy, ion-oxygen distance, and coordination number in the first hydration shell accurately for most of the metal cations, an accuracy equivalent to that of the complex 12-6-4 LJ-type and double exponential potential models. A systematic comparison with the existing M²⁺ models is presented as well in terms of effective ion radii, diffusion constants, water exchange rates, and ion-water interactions. Molecular dynamics simulations of metal substitution in Escherichia coli glyoxalase I variants show the great potential of our new models for metalloproteins.

Evaluation of nine condensed-phase force fields of the GROMOS, CHARMM, OPLS, AMBER, and OpenFF families against experimental cross-solvation free energies

Experimental solvation free energies are nowadays commonly included as target properties in the validation of condensed-phase force fields, sometimes even in their calibration. In a previous article [Kashefolgheta et al., J. Chem. Theory. Comput., 2020, 16, 7556-7580], we showed how the involved comparison between experimental and simulation results could be made more systematic by considering a full matrix of cross-solvation free energies . For a set of N molecules that are all in the liquid state under ambient conditions, such a matrix encompasses N×N entries for considering each of the N molecules either as solute (A) or as solvent (B). In the quoted study, a cross-solvation matrix of 25 × 25 experimental value was introduced, considering 25 small molecules representative for alkanes, chloroalkanes, ethers, ketones, esters, alcohols, amines, and amides. This experimental data was used to compare the relative accuracies of four popular condensed-phase force fields, namely GROMOS-2016H66, OPLS-AA, AMBER-GAFF, and CHARMM-CGenFF. In the present work, the comparison is extended to five additional force fields, namely GROMOS-54A7, GROMOS-ATB, OPLS-LBCC, AMBER-GAFF2, and OpenFF. Considering these nine force fields, the correlation coefficients between experimental values and simulation results range from 0.76 to 0.88, the root-mean-square errors (RMSEs) from 2.9 to 4.8 kJ mol-1, and average errors (AVEEs) from -1.5 to +1.0 kJ mol-1. In terms of RMSEs, GROMOS-2016H66 and OPLS-AA present the best accuracy (2.9 kJ mol-1), followed by OPLS-LBCC, AMBER-GAFF2, AMBER-GAFF, and OpenFF (3.3 to 3.6 kJ mol-1), and then by GROMOS-54A7, CHARM-CGenFF, and GROMOS-ATB (4.0 to 4.8 kJ mol-1). These differences are statistically significant but not very pronounced, and are distributed rather heterogeneously over the set of compounds within the different force fields.

A fast and high-quality charge model for the next generation general AMBER force field

The General AMBER Force Field (GAFF) has been broadly used by researchers all over the world to perform in silico simulations and modelings on diverse scientific topics, especially in the field of computer-aided drug design whose primary task is to accurately predict the affinity and selectivity of receptor-ligand binding. The atomic partial charges in GAFF and the second generation of GAFF (GAFF2) were originally developed with the quantum mechanics derived restrained electrostatic potential charge, but in practice, users usually adopt an efficient charge method, Austin Model 1-bond charge corrections (AM1-BCC), based on which, without expensive ab initio calculations, the atomic charges could be efficiently and conveniently obtained with the ANTECHAMBER module implemented in the AMBER software package. In this work, we developed a new set of BCC parameters specifically for GAFF2 using 442 neutral organic solutes covering diverse functional groups in aqueous solution. Compared to the original BCC parameter set, the new parameter set significantly reduced the mean unsigned error (MUE) of hydration free energies from 1.03 kcal/mol to 0.37 kcal/mol. More excitingly, this new AM1-BCC model also showed excellent performance in the solvation free energy (SFE) calculation on diverse solutes in various organic solvents across a range of different dielectric constants. In this large-scale test with totally 895 neutral organic solvent-solute systems, the new parameter set led to accurate SFE predictions with the MUE and the root-mean-square-error of 0.51 kcal/mol and 0.65 kcal/mol, respectively. This newly developed charge model, ABCG2, paved a promising path for the next generation GAFF development.

Systematic design of biomolecular force fields

Force fields for the study of biomolecules have been developed in a predominantly organic manner by regular updates over half a century. Together with better algorithms and advances in computer technology, force fields have improved to yield more robust predictions. However, there are also indications to suggest that intramolecular energy functions have not become better and that there still is room for improvement. In this review, systematic efforts toward development of novel force fields from scratch are described. This includes an estimate of the complexity of the problem and the prerequisites in the form of data and algorithms. It is suggested that in order to make progress, an effort is needed to standardize reference data and force field validation benchmarks.

A potential for molecular simulation of compounds with linear moieties

The harmonic angle bending potential is used in many force fields for (bio)molecular simulation. The force associated with this potential is discontinuous at angles close to 180°, which can lead to numeric instabilities. Angle bending of linear groups, such as alkynes or nitriles, or linear molecules, such as carbon dioxide, can be treated by a simple harmonic potential if we describe the fluctuations as a deviation from a reference position of the central atom, the position of which is determined by the flanking atoms. The force constant for the linear angle potential can be derived analytically from the corresponding force constant in the traditional potential. The new potential is tested on the properties of alkynes, nitriles, and carbon dioxide. We find that the angles of the linear groups remain about 2° closer to 180° using the new potential. The bond and angle force constants for carbon dioxide were tuned to reproduce the experimentally determined frequencies. An interesting finding was that simulations of liquid carbon dioxide under pressure with the new flexible model were stable only when explicitly modeling the long-range Lennard-Jones (LJ) interactions due to the very long-range nature of the LJ interactions (>1.7 nm). In the other tested liquids, we find that a Lennard-Jones cutoff of 1.1 nm yields similar results as the particle mesh Ewald algorithm for LJ interactions. Algorithmic factors influencing the stability of liquid simulations are discussed as well. Finally, we demonstrate that the linear angle potential can be used in free energy perturbation calculations.

Theoretical Infrared Spectra: Quantitative Similarity Measures and Force Fields

Infrared spectroscopy can provide significant insight into the structures and dynamics of molecules of all sizes. The information that is contained in the spectrum is, however, often not easily extracted without the aid of theoretical calculations or simulations. We present here the calculation of the infrared spectra of a database of 703 gas phase compounds with four different force fields (CGenFF, GAFF-BCC, GAFF-ESP, and OPLS) using normal-mode analysis. Modern force fields increasingly use virtual sites to describe, e.g., lone-pair electrons or the σ-holes on halogen atoms. This requires some adaptation of code to perform normal-mode analysis of such compounds, the implementation of which into the GROMACS software is briefly described as well. For the quantitative comparison of the obtained spectra with experimental reference data, we discuss the application of two different statistical correlation coefficients, Pearson and Spearman. The advantages and drawbacks of the different methods of comparison are discussed, and we find that both methods of comparison give the same overall picture, showing that present force field methods cannot match the performance of quantum chemical methods for the calculation of infrared spectra.

Rotational and Translational Diffusion of Proteins as a Function of Concentration

Atomistic simulations of three different proteins at different concentrations are performed to obtain insight into protein mobility as a function of protein concentration. We report on simulations of proteins from diluted to the physiological water concentration (about 70% of the mass). First, the viscosity was computed and found to increase by a factor of 7-9 going from pure water to the highest protein concentration, in excellent agreement with in vivo nuclear magnetic resonance results. At a physiological concentration of proteins, the translational diffusion is found to be slowed down to about 30% of the in vitro values. The slow-down of diffusion found here using atomistic models is slightly more than that of a hard sphere model that neglects the electrostatic interactions. Interestingly, rotational diffusion of proteins is slowed down somewhat more (by about 80-95% compared to in vitro values) than translational diffusion, in line with experimental findings and consistent with the increased viscosity. The finding that rotation is retarded more than translation is attributed to solvent-separated clustering. No direct interactions between the proteins are found, and the clustering can likely be attributed to dispersion interactions that are stronger between proteins than between protein and water. Based on these simulations, we can also conclude that the internal dynamics of the proteins in our study are affected only marginally under crowding conditions, and the proteins become somewhat more stable at higher concentrations. Simulations were performed using a force field that was tuned for dealing with crowding conditions by strengthening the protein-water interactions. This force field seems to lead to a reproducible partial unfolding of an α-helix in one of the proteins, an effect that was not observed in the unmodified force field.

Solvent-Mediated Affinity of Polymer-Wrapped Single-Walled Carbon Nanotubes for Chemically Modified Surfaces

Semiconducting single-walled carbon nanotube (s-CNT) arrays are being explored for next-generation semiconductor electronics. Even with the multitude of alignment and spatially localized s-CNT deposition methods designed to control s-CNT deposition, fundamental understanding of the driving forces for s-CNT deposition is still lacking. The individual roles of the dispersant, solvent, target substrate composition, and the s-CNT itself are not completely understood because it is difficult to decouple deposition parameters. Here, we study poly[(9,9-dioctylfluorenyl-2,7-diyl)-alt-co-(6,6'-[2,2'-{bipyridine}])] (PFO-BPy)-wrapped s-CNT deposition from solution onto a chemically modified substrate. We fabricate various self-assembled monolayers (SAMs) to gain a greater understanding of substrate effects on PFO-BPy-wrapped s-CNT deposition. We observe that s-CNT deposition is dependent on both the target substrate and s-CNT dispersion solvent. To complement the experiments, molecular dynamics simulations of PFO-BPy-wrapped s-CNT deposition on two different SAMs are performed to obtain mechanistic insights into the effect of the substrate and solvent on s-CNT deposition. We find that the global free-energy minimum associated with favorable s-CNT adsorption occurs for a configuration in which the minimum of the solvent density around the s-CNT coincides with the minimum of the solvent density above a SAM-grafted surface, indicating that solvent structure near a SAM-grafted surface determines the adsorption free-energy landscape driving s-CNT deposition. Our results will help guide informative substrate design for s-CNT array fabrication in semiconductor devices.

Prediction of Partition Coefficients of Environmental Toxins Using Computational Chemistry Methods

The partitioning of compounds between aqueous and other phases is important for predicting toxicity. Although thousands of octanol-water partition coefficients have been measured, these represent only a small fraction of the anthropogenic compounds present in the environment. The octanol phase is often taken to be a mimic of the inner parts of phospholipid membranes. However, the core of such membranes is typically more hydrophobic than octanol, and other partition coefficients with other compounds may give complementary information. Although a number of (cheap) empirical methods exist to compute octanol-water (log k _OW) and hexadecane-water (log k _HW) partition coefficients, it would be interesting to know whether physics-based models can predict these crucial values more accurately. Here, we have computed log k _OW and log k _HW for 133 compounds from seven different pollutant categories as well as a control group using the solvation model based on electronic density (SMD) protocol based on Hartree-Fock (HF) or density functional theory (DFT) and the COSMO-RS method. For comparison, XlogP3 (log k _OW) values were retrieved from the PubChem database, and KowWin log k _OW values were determined as well. For 24 of these compounds, log k _OW was computed using potential of mean force (PMF) calculations based on classical molecular dynamics simulations. A comparison of the accuracy of the methods shows that COSMO-RS, KowWin, and XlogP3 all have a root-mean-square deviation (rmsd) from the experimental data of ≈0.4 log units, whereas the SMD protocol has an rmsd of 1.0 log units using HF and 0.9 using DFT. PMF calculations yield the poorest accuracy (rmsd = 1.1 log units). Thirty-six out of 133 calculations are for compounds without known log k _OW, and for these, we provide what we consider a robust prediction, in the sense that there are few outliers, by averaging over the methods. The results supplied may be instrumental when developing new methods in computational ecotoxicity. The log k _HW values are found to be strongly correlated to log k _OW for most compounds.

Concepts of Artificial Intelligence for Computer-Assisted Drug Discovery

Artificial intelligence (AI), and, in particular, deep learning as a subcategory of AI, provides opportunities for the discovery and development of innovative drugs. Various machine learning approaches have recently (re)emerged, some of which may be considered instances of domain-specific AI which have been successfully employed for drug discovery and design. This review provides a comprehensive portrayal of these machine learning techniques and of their applications in medicinal chemistry. After introducing the basic principles, alongside some application notes, of the various machine learning algorithms, the current state-of-the art of AI-assisted pharmaceutical discovery is discussed, including applications in structure- and ligand-based virtual screening, de novo drug design, physicochemical and pharmacokinetic property prediction, drug repurposing, and related aspects. Finally, several challenges and limitations of the current methods are summarized, with a view to potential future directions for AI-assisted drug discovery and design.

COSMOplex: self-consistent simulation of self-organizing inhomogeneous systems based on COSMO-RS

During the past 20 years, the efficient combination of quantum chemical calculations with statistical thermodynamics by the COSMO-RS method has become an important alternative to force-field based simulations for the accurate prediction of free energies of molecules in liquid systems. While it was originally restricted to homogeneous liquids, it later has been extended to the prediction of the free energy of molecules in inhomogeneous systems such as micelles, biomembranes, or liquid interfaces, but these calculations were based on external input about the structure of the inhomogeneous system. Here we report the rigorous extension of COSMO-RS to a self-consistent prediction of the structure and the free energies of molecules in self-organizing inhomogeneous systems. This extends the application range to many new areas, such as the prediction of micellar structures and critical micelle concentrations, finite loading effects in micelles and biomembranes, the free energies and structure of liquid interfaces, microemulsions, and many more related topics, which often are of great practical importance.

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.

In situ monitoring of molecular aggregation using circular dichroism

The aggregation of molecules plays an important role in determining their function. Electron microscopy and other methods can only characterize the variation of microstructure, but are not capable of monitoring conformational changes. These techniques are also complicated, expensive and time-consuming. Here, we demonstrate a simple method to monitor in-situ and in real-time the conformational change of (R)-1,1'-binaphthyl-based polymers during the aggregation process using circular dichroism. Based on results from molecular dynamics simulations and experimental circular dichroism measurements, polymers with "open" binaphthyl rings are found to show stronger aggregation-annihilated circular dichroism effects, with more negative torsion angles between the two naphthalene rings. In contrast, the polymers with "locked" rings show a more restrained aggregation-annihilated circular dichroism effect, with only a slight change of torsion angle. This work provides an approach to monitor molecular aggregation in a simple, accurate, and efficient way.

Solvation free energies and partition coefficients with the coarse-grained and hybrid all-atom/coarse-grained MARTINI models

We present the estimation of solvation free energies of small solutes in water, n-octanol and hexane using molecular dynamics simulations with two MARTINI models at different resolutions, viz. the coarse-grained (CG) and the hybrid all-atom/coarse-grained (AA/CG) models. From these estimates, we also calculate the water/hexane and water/octanol partition coefficients. More than 150 small, organic molecules were selected from the Minnesota solvation database and parameterized in a semi-automatic fashion. Using either the CG or hybrid AA/CG models, we find considerable deviations between the estimated and experimental solvation free energies in all solvents with mean absolute deviations larger than 10 kJ/mol, although the correlation coefficient is between 0.55 and 0.75 and significant. There is also no difference between the results when using the non-polarizable and polarizable water model, although we identify some improvements when using the polarizable model with the AA/CG solutes. In contrast to the estimated solvation energies, the estimated partition coefficients are generally excellent with both the CG and hybrid AA/CG models, giving mean absolute deviations between 0.67 and 0.90 log units and correlation coefficients larger than 0.85. We analyze the error distribution further and suggest avenues for improvements.

Approaches for calculating solvation free energies and enthalpies demonstrated with an update of the FreeSolv database

Solvation free energies can now be calculated precisely from molecular simulations, providing a valuable test of the energy functions underlying these simulations. Here, we briefly review "alchemical" approaches for calculating the solvation free energies of small, neutral organic molecules from molecular simulations, and illustrate by applying them to calculate aqueous solvation free energies (hydration free energies). These approaches use a non-physical pathway to compute free energy differences from a simulation or set of simulations and appear to be a particularly robust and general-purpose approach for this task. We also present an update (version 0.5) to our FreeSolv database of experimental and calculated hydration free energies of neutral compounds and provide input files in formats for several simulation packages. This revision to FreeSolv provides calculated values generated with a single protocol and software version, rather than the heterogeneous protocols used in the prior version of the database. We also further update the database to provide calculated enthalpies and entropies of hydration and some experimental enthalpies and entropies, as well as electrostatic and nonpolar components of solvation free energies.