Testing the semi-explicit assembly model of aqueous solvation in the SAMPL4 challenge

Here, we test a method, called semi-explicit assembly (SEA), that computes the solvation free energies of molecules in water in the SAMPL4 blind test challenge. SEA was developed with the intention of being as accurate as explicit-solvent models, but much faster to compute. It is accurate because it uses pre-simulations of simple spheres in explicit solvent to obtain structural and thermodynamic quantities, and it is fast because it parses solute free energies into regionally additive quantities. SAMPL4 provided us the opportunity to make new tests of SEA. Our tests here lead us to the following conclusions: (1) The newest version, called Field-SEA, which gives improved predictions for highly charged ions, is shown here to perform as well as the earlier versions (dipolar and quadrupolar SEA) on this broad blind SAMPL4 test set. (2) We find that both the past and present SEA models give solvation free energies that are as accurate as TIP3P. (3) Using a new approach for force field parameter optimization, we developed improved hydroxyl parameters that ensure consistency with neat-solvent dielectric constants, and found that they led to improved solvation free energies for hydroxyl-containing compounds in SAMPL4. We also learned that these hydroxyl parameters are not just fixing solvent exposed oxygens in a general sense, and therefore do not improve predictions for carbonyl or carboxylic-acid groups. Other such functional groups will need their own independent optimizations for potential improvements. Overall, these tests in SAMPL4 indicate that SEA is an accurate, general and fast new approach to computing solvation free energies.     

Testing the semi-explicit assembly solvation model in the SAMPL3 community blind test

We report here a test of the Semi-Explicit Assembly (SEA) model in the solvation free energy category of the SAMPL3 blind prediction event (summer 2011). We tested how dependent the SEA results are on the chosen force field by performing calculations with both the General Amber and OPLS force fields. We compared our SEA results with full molecular dynamics simulations in explicit solvent. Of the 20 submissions, our SEA/OPLS results gave the second smallest RMS errors in free energies compared to experiments. SEA gives results that are very similar to those of its underlying force field and explicit solvent model. Hence, while the SEA water modeling approach is much faster than explicit solvent simulations, its predictions appear to be just as accurate.     

SAMPL5: 3D-RISM partition coefficient calculations with partial molar volume corrections and solute conformational sampling

Implicit solvent methods for classical molecular modeling are frequently used to provide fast, physics-based hydration free energies of macromolecules. Less commonly considered is the transferability of these methods to other solvents. The Statistical Assessment of Modeling of Proteins and Ligands 5 (SAMPL5) distribution coefficient dataset and the accompanying explicit solvent partition coefficient reference calculations provide a direct test of solvent model transferability. Here we use the 3D reference interaction site model (3D-RISM) statistical-mechanical solvation theory, with a well tested water model and a new united atom cyclohexane model, to calculate partition coefficients for the SAMPL5 dataset. The cyclohexane model performed well in training and testing (\(R=0.98\) for amino acid neutral side chain analogues) but only if a parameterized solvation free energy correction was used. In contrast, the same protocol, using single solute conformations, performed poorly on the SAMPL5 dataset, obtaining \(R=0.73\) compared to the reference partition coefficients, likely due to the much larger solute sizes. Including solute conformational sampling through molecular dynamics coupled with 3D-RISM (MD/3D-RISM) improved agreement with the reference calculation to \(R=0.93\). Since our initial calculations only considered partition coefficients and not distribution coefficients, solute sampling provided little benefit comparing against experiment, where ionized and tautomer states are more important. Applying a simple \(\hbox {p}K_{\text {a}}\) correction improved agreement with experiment from \(R=0.54\) to \(R=0.66\), despite a small number of outliers. Better agreement is possible by accounting for tautomers and improving the ionization correction.     

Prediction of SAMPL2 aqueous solvation free energies and tautomeric ratios using the SM8, SM8AD,and SMD solvation models

We applied the solvation models SM8, SM8AD, and SMD in combination with the Minnesota M06-2X density functional to predict vacuum-water transfer free energies (Task 1) and tautomeric ratios in aqueous solution (Task 2) for the SAMPL2 test set. The bulk-electrostatic contribution to the free energy of solvation is treated as follows: SM8 employs the generalized Born model with the Coulomb field approximation, SM8AD employs the generalized Born approximation with asymmetric descreening, and SMD solves the nonhomogeneous Poisson equation. The non-bulk-electrostatic contribution arising from short-range interactions between the solute and solvent molecules in the first solvation shell is treated as a sum of terms that are products of geometry-dependent atomic surface tensions and solvent-accessible surface areas of the individual atoms of the solute. On average, three models tested in the present work perform similarly. In particular, we achieved mean unsigned errors of 1.3 (SM8), 2.0 (SM8AD), and 2.6 kcal/mol (SMD) for the aqueous free energies of 30 out of 31 compounds with known reference data involved in Task 1 and mean unsigned errors of 2.7 (SM8), 1.8 (SM8AD), and 2.4 kcal/mol (SMD) in the free energy differences (tautomeric ratios) for 21 tautomeric pairs in aqueous solution involved in Task 2.     

Predicting water-to-cyclohexane partitioning of the SAMPL5 molecules using dielectric balancing of force fields

Alchemical transformation of solutes using classical fixed-charge force fields is a popular strategy for assessing the free energy of transfer in different environments. Accurate estimations of transfer between phases with significantly different polarities can be difficult because of the static nature of the force fields. Here, we report on an application of such calculations in the SAMPL5 experiment that also involves an effort in balancing solute and solvent interactions via their expected static dielectric constants. This strategy performs well with respect to predictive accuracy and correlation with unknown experimental values. We follow this by performing a series of retrospective investigations which highlight the potential importance of proper balancing in these systems, and we use a null hypothesis analysis to explore potential biases in the comparisons with experiment. The collective findings indicate that considerations of force field compatibility through dielectric behavior is a potential strategy for future improvements in transfer processes between disparate environments.     

Rapid prediction of solvation free energy. 3. Application to the SAMPL2 challenge

The SAMPL2 hydration free energy blind prediction challenge consisted of a data set of 41 molecules divided into three subsets: explanatory, obscure and investigatory, where experimental hydration free energies were given for the explanatory, withheld for the obscure, and not known for the investigatory molecules. We employed two solvation models for this challenge, a linear interaction energy (LIE) model based on explicit-water molecular dynamics simulations, and the first-shell hydration (FiSH) continuum model previously calibrated to mimic LIE data. On the 23 compounds from the obscure (blind) dataset, the prospectively submitted LIE and FiSH models provided predictions highly correlated with experimental hydration free energy data, with mean-unsigned-errors of 1.69 and 1.71 kcal/mol, respectively. We investigated several parameters that may affect the performance of these models, namely, the solute flexibility for the LIE explicit-solvent model, the solute partial charging method, and the incorporation of the difference in intramolecular energy between gas and solution phases for both models. We extended this analysis to the various chemical classes that can be formed within the SAMPL2 dataset. Our results strengthen previous findings on the excellent accuracy and transferability of the LIE explicit-solvent approach to predict transfer free energies across a wide spectrum of functional classes. Further, the current results on the SAMPL2 test dataset provide additional support for the FiSH continuum model as a fast yet accurate alternative to the LIE explicit-solvent model. Overall, both the LIE explicit-solvent model and the FiSH continuum solvation model show considerable improvement on the SAMPL2 data set over our previous continuum electrostatics-dispersion solvation model used in the SAMPL1 blind challenge.     

Blind prediction of SAMPL4 cucurbit[7]uril binding affinities with the mining minima method

Accurate methods for predicting protein–ligand binding affinities are of central interest to computer-aided drug design for hit identification and lead optimization. Here, we used the mining minima (M2) method to predict cucurbit[7]uril binding affinities from the SAMPL4 blind prediction challenge. We tested two different energy models, an empirical classical force field, CHARMm with VCharge charges, and the Poisson–Boltzmann surface area solvation model; and a semiempirical quantum mechanical (QM) Hamiltonian, PM6-DH+, coupled with the COSMO solvation model and a surface area term for nonpolar solvation free energy. Binding affinities based on the classical force field correlated strongly with the experiments with a correlation coefficient (R²) of 0.74. On the other hand, binding affinities based on the QM energy model correlated poorly with experiments (R² = 0.24), due largely to two major outliers. As we used extensive conformational search methods, these results point to possible inaccuracies in the PM6-DH+ energy model or the COSMO solvation model. Furthermore, the different binding free energy components, solute energy, solvation free energy, and configurational entropy showed significant deviations between the classical M2 and quantum M2 calculations. Comparison of different classical M2 free energy components to experiments show that the change in the total energy, i.e. the solute energy plus the solvation free energy, is the key driving force for binding, with a reasonable correlation to experiment (R² = 0.56); however, accounting for configurational entropy further improves the correlation.     

Predicting hydration free energies using all-atom molecular dynamics simulations and multiple starting conformations

Molecular dynamics simulations in explicit solvent were applied to predict the hydration free energies for 23 small organic molecules in blind SAMPL2 test. We found good agreement with experimental results, with an RMS error of 2.82 kcal/mol over the whole set and 1.86 kcal/mol over all the molecules except several hydroxyl-rich compounds where we find evidence for a systematic error in the force field. We tested two different solvent models, TIP3P and TIP4P-Ew, and obtained very similar hydration free energies for these two models; the RMS difference was 0.64 kcal/mol. We found that preferred conformation of the carboxylic acids in water differs from that in vacuum. Surprisingly, this conformational change is not adequately sampled on simulation timescales, so we apply an umbrella sampling technique to include free energies associated with the conformational change. Overall, the results of this test reveal that the force field parameters for some groups of molecules (such as hydroxyl-rich compounds) still need to be improved, but for most compounds, accuracy was consistent with that seen in our previous tests.     

Blind prediction of cyclohexane–water distribution coefficients from the SAMPL5 challenge

In the recent SAMPL5 challenge, participants submitted predictions for cyclohexane/water distribution coefficients for a set of 53 small molecules. Distribution coefficients (log D) replace the hydration free energies that were a central part of the past five SAMPL challenges. A wide variety of computational methods were represented by the 76 submissions from 18 participating groups. Here, we analyze submissions by a variety of error metrics and provide details for a number of reference calculations we performed. As in the SAMPL4 challenge, we assessed the ability of participants to evaluate not just their statistical uncertainty, but their model uncertainty—how well they can predict the magnitude of their model or force field error for specific predictions. Unfortunately, this remains an area where prediction and analysis need improvement. In SAMPL4 the top performing submissions achieved a root-mean-squared error (RMSE) around 1.5 kcal/mol. If we anticipate accuracy in log D predictions to be similar to the hydration free energy predictions in SAMPL4, the expected error here would be around 1.54 log units. Only a few submissions had an RMSE below 2.5 log units in their predicted log D values. However, distribution coefficients introduced complexities not present in past SAMPL challenges, including tautomer enumeration, that are likely to be important in predicting biomolecular properties of interest to drug discovery, therefore some decrease in accuracy would be expected. Overall, the SAMPL5 distribution coefficient challenge provided great insight into the importance of modeling a variety of physical effects. We believe these types of measurements will be a promising source of data for future blind challenges, especially in view of the relatively straightforward nature of the experiments and the level of insight provided.     

Lessons learned from comparing molecular dynamics engines on the SAMPL5 dataset

We describe our efforts to prepare common starting structures and models for the SAMPL5 blind prediction challenge. We generated the starting input files and single configuration potential energies for the host-guest in the SAMPL5 blind prediction challenge for the GROMACS, AMBER, LAMMPS, DESMOND and CHARMM molecular simulation programs. All conversions were fully automated from the originally prepared AMBER input files using a combination of the ParmEd and InterMol conversion programs. We find that the energy calculations for all molecular dynamics engines for this molecular set agree to better than 0.1 % relative absolute energy for all energy components, and in most cases an order of magnitude better, when reasonable choices are made for different cutoff parameters. However, there are some surprising sources of statistically significant differences. Most importantly, different choices of Coulomb’s constant between programs are one of the largest sources of discrepancies in energies. We discuss the measures required to get good agreement in the energies for equivalent starting configurations between the simulation programs, and the energy differences that occur when simulations are run with program-specific default simulation parameter values. Finally, we discuss what was required to automate this conversion and comparison.     

Prediction of cyclohexane-water distribution coefficient for SAMPL5 drug-like compounds with the QMPFF3 and ARROW polarizable force fields

We present the performance of blind predictions of water—cyclohexane distribution coefficients for 53 drug-like compounds in the SAMPL5 challenge by three methods currently in use within our group. Two of them utilize QMPFF3 and ARROW, polarizable force-fields of varying complexity, and the third uses the General Amber Force-Field (GAFF). The polarizable FF’s are implemented in an in-house MD package, Arbalest. We find that when we had time to parametrize the functional groups with care (batch 0), the polarizable force-fields outperformed the non-polarizable one. Conversely, on the full set of 53 compounds, GAFF performed better than both QMPFF3 and ARROW. We also describe the torsion-restrain method we used to improve sampling of molecular conformational space and thus the overall accuracy of prediction. The SAMPL5 challenge highlighted several drawbacks of our force-fields, such as our significant systematic over-estimation of hydrophobic interactions, specifically for alkanes and aromatic rings.     

The SAMPL6 challenge on predicting aqueous p<Emphasis Type="Italic">K</Emphasis><Subscript>a</Subscript> values from EC-RISM theory

The “embedded cluster reference interaction site model” (EC-RISM) integral equation theory is applied to the problem of predicting aqueous pK_a values for drug-like molecules based on an ensemble of tautomers. EC-RISM is based on self-consistent calculations of a solute’s electronic structure and the distribution function of surrounding water. Following-up on the workflow developed after the SAMPL5 challenge on cyclohexane-water distribution coefficients we extended and improved the methodology by taking into account exact electrostatic solute–solvent interactions taken from the wave function in solution. As before, the model is calibrated against Gibbs energies of hydration from the “Minnesota Solvation Database” and a public dataset of acidity constants of organic acids and bases by adjusting in total 4 parameters, among which only 3 are relevant for predicting pK_a values. While the best-performing training model yields a root-mean-square error (RMSE) of 1 pK unit, the corresponding test set prediction on the full SAMPL6 dataset of macroscopic pK_a values using the same level of theory exhibits slightly larger error (1.7 pK units) than the best test set model submitted (1.7 pK units for corresponding training set vs. test set performance of 1.6). Post-submission analysis revealed a number of physical optimization options regarding the numerical treatment of electrostatic interactions and conformational sampling. While the experimental test set data revealed after submission was not used for reparametrizing the methodology, the best physically optimized models consequentially result in RMSEs of 1.5 if only improved electrostatic interactions are considered and of 1.1 if, in addition, conformational sampling accounts for quantum-chemically derived rankings. We conclude that these numbers are probably near the ultimate accuracy achievable with the simple 3-parameter model using a single or the two best-ranking conformations per tautomer or microstate. Finally, relations of the present macrostate approach to microstate pK_a results are discussed and some illustrative results for microstate populations are presented.     

A new approach to the determination of concentration profiles in atom probe tomography

Atom probe tomography (APT) provides three-dimensional analytical imaging of materials with near-atomic resolution using pulsed field evaporation. The processes of field evaporation can cause atoms to be placed at positions in the APT reconstruction that can deviate slightly from their original site in the material. Here, we describe and model one such process--that of preferential retention of solute atoms in multicomponent systems. Based on relative field evaporation probabilities, we calculate the point spread function for the solute atom distribution in the "z," or in-depth direction, and use this to extract more accurate solute concentration profiles.     

Effect of the Solute Cavity on the Solvation Energy and its Derivatives within the Framework of the Gaussian Charge Scheme

The treatment of the solvation charges using Gaussian functions in the polarizable continuum model results in a smooth potential energy surface. These charges are placed on top of the surface of the solute cavity. In this article, we study the effect of the solute cavity (van der Waals-type or solvent-excluded surface-type) using the Gaussian charge scheme within the framework of the conductor-like polarizable continuum model on (a) the accuracy and computational cost of the self-consistent field (SCF) energy and its gradient and on (b) the calculation of free energies of solvation. For that purpose, we have considered a large set of systems ranging from few atoms to more than 200 atoms in different solvents. Our results at the DFT level using the B3LYP functional and the def2-TZVP basis set show that the choice of the solute cavity does neither affect the accuracy nor the cost of calculations for small systems (< 100 atoms). For larger systems, the use of a vdW-type cavity is recommended, as it prevents small oscillations in the gradient (present when using a SES-type cavity), which affect the convergence of the SCF energy gradient. Regarding the free energies of solvation, we consider a solvent-dependent probe sphere to construct the solvent-accessible surface area required to calculate the nonelectrostatic contribution to the free energy of solvation. For this part, our results for a large set of organic molecules in different solvents agree with available experimental data with an accuracy lower than 1 kcal/mol for both polar and nonpolar solvents.     

Extensive all-atom Monte Carlo sampling and QM/MM corrections in the SAMPL4 hydration free energy challenge

We present our predictions for the SAMPL4 hydration free energy challenge. Extensive all-atom Monte Carlo simulations were employed to sample the compounds in explicit solvent. While the focus of our study was to demonstrate well-converged and reproducible free energies, we attempted to address the deficiencies in the general Amber force field force field with a simple QM/MM correction. We show that by using multiple independent simulations, including different starting configurations, and enhanced sampling with parallel tempering, we can obtain well converged hydration free energies. Additional analysis using dihedral angle distributions, torsion-root mean square deviation plots and thermodynamic cycles support this assertion. We obtain a mean absolute deviation of 1.7 kcal mol^?1 and a Kendall’s τ of 0.65 compared with experiment.     

Enumerating the total colorings of a polyhedron and application to polyhedral links

In this paper we consider the enumeration problem of a particular three-dimensional molecular or chemical compound system which has a polyhedral frame where the vertices, edges and faces represent ‘units’ such as atoms, bonds, ligands, polymers, or other objects of chemical interests. In this system, chirality is also taken into account. This enumeration problem is mathematically modeled as the ‘total coloring’ enumeration problem of a polyhedron: i.e., the number of ways to color all the vertices, edges and faces of the polyhedron by using three or more corresponding color sets, in which some colors may be chiral. We establish a general formula for this enumeration problem by extending the fundamental version of Pólya’s enumeration theorem. In particular, we apply this technique to the enumeration problem of polyhedral links which have received special attention from biochemists, mathematical chemists and mathematicians over the past two decades.     

Application of thermal analysis to carbon black-elastomer systems

In this study a relationship between the surface and textural properties of carbon blacks and the gasification process induced thermally in air is noted. A temperature jump method was used to follow the gasification in air and to establish the relevant Arrhenius parameters. This can be associated with the activity of ‘basal’ plane carbon atoms as well as ‘edge’ carbon atoms at the surface. This is based on a model of carbon black structure consisting of the irregular packing of small graphite carbon structures. The carbon black surface was measured using a single measurement of adsorption based essentially on the BET volumetric method. The carbon black surface had a rate of oxidation per unit are which clearly indicated that the lower area carbon blacks had a predominately active area of ‘edge’ atoms at the surface while the surface of the higher area carbon blacks had a predominate amount of ‘basal’ plane graphite carbon atoms.     

Breaking the polar‐nonpolar division in solvation free energy prediction

Implicit solvent models divide solvation free energies into polar and nonpolar additive contributions, whereas polar and nonpolar interactions are inseparable and nonadditive. We present a feature functional theory (FFT) framework to break this ad hoc division. The essential ideas of FFT are as follows: (i) representability assumption: there exists a microscopic feature vector that can uniquely characterize and distinguish one molecule from another; (ii) feature‐function relationship assumption: the macroscopic features, including solvation free energy, of a molecule is a functional of microscopic feature vectors; and (iii) similarity assumption: molecules with similar microscopic features have similar macroscopic properties, such as solvation free energies. Based on these assumptions, solvation free energy prediction is carried out in the following protocol. First, we construct a molecular microscopic feature vector that is efficient in characterizing the solvation process using quantum mechanics and Poisson–Boltzmann theory. Microscopic feature vectors are combined with macroscopic features, that is, physical observable, to form extended feature vectors. Additionally, we partition a solvation dataset into queries according to molecular compositions. Moreover, for each target molecule, we adopt a machine learning algorithm for its nearest neighbor search, based on the selected microscopic feature vectors. Finally, from the extended feature vectors of obtained nearest neighbors, we construct a functional of solvation free energy, which is employed to predict the solvation free energy of the target molecule. The proposed FFT model has been extensively validated via a large dataset of 668 molecules. The leave‐one‐out test gives an optimal root‐mean‐square error (RMSE) of 1.05 kcal/mol. FFT predictions of SAMPL0, SAMPL1, SAMPL2, SAMPL3, and SAMPL4 challenge sets deliver the RMSEs of 0.61, 1.86, 1.64, 0.86, and 1.14 kcal/mol, respectively. Using a test set of 94 molecules and its associated training set, the present approach was carefully compared with a classic solvation model based on weighted solvent accessible surface area. © 2017 Wiley Periodicals, Inc.     

Atoms of multistationarity in chemical reaction networks

Chemical reaction systems are dynamical systems that arise in chemical engineering and systems biology. In this work, we consider the question of whether the minimal (in a precise sense) multistationary chemical reaction networks, which we propose to call ‘atoms of multistationarity,’ characterize the entire set of multistationary networks. Our main result states that the answer to this question is ‘yes’ in the context of fully open continuous-flow stirred-tank reactors (CFSTRs), which are networks in which all chemical species take part in the inflow and outflow. In order to prove this result, we show that if a subnetwork admits multiple steady states, then these steady states can be lifted to a larger network, provided that the two networks share the same stoichiometric subspace. We also prove an analogous result when a smaller network is obtained from a larger network by ‘removing species.’ Our results provide the mathematical foundation for a technique used by Siegal- Gaskins et al. of establishing bistability by way of ‘network ancestry.’ Additionally, our work provides sufficient conditions for establishing multistationarity by way of atoms and moreover reduces the problem of classifying multistationary CFSTRs to that of cataloging atoms of multistationarity. As an application, we enumerate and classify all 386 bimolecular and reversible two-reaction networks. Of these, exactly 35 admit multiple positive steady states. Moreover, each admits a unique minimal multistationary subnetwork, and these subnetworks form a poset (with respect to the relation of ‘removing species’) which has 11 minimal elements (the atoms of multistationarity).