Predicting hydration free energies with chemical accuracy: the SAMPL4 challenge

An implicit solvent model described by a non-simple dielectric medium is used for the prediction of hydration free energies on the dataset of 47 molecules in the SAMPL4 challenge. The solute is represented by a minimal parameter set model based on a new all atom force-field, named the liquid simulation force-field. The importance of a first solvation shell correction to the hydration free energy prediction is discussed and two different approaches are introduced to address it: either with an empirical correction to a few functional groups (alcohol, ether, ester, amines and aromatic nitrogen), or an ab initio correction based on the formation of a solute/explicit water complex. Both approaches give equally good predictions with an average unsigned error <1 kcal/mol. Chemical accuracy is obtained.     

Prediction of hydration free energies for the SAMPL4 diverse set of compounds using molecular dynamics simulations with the OPLS-AA force field

All-atom molecular dynamics computer simulations were used to blindly predict the hydration free energies of a range of small molecules as part of the SAMPL4 challenge. Compounds were parametrized on the basis of the OPLS-AA force field using three different protocols for deriving partial charges: (1) using existing OPLS-AA atom types and charges with minor adjustments of partial charges on equivalent connecting atoms and derivation of new parameters for a number of distinct chemical groups (N-alkyl imidazole, nitrate) that were not present in the published force field; (2) calculation of quantum mechanical charges via geometry optimization, followed by electrostatic potential (ESP) fitting, using Jaguar at the LMP2/cc-pVTZ(-F) level; and (3) via geometry optimization and CHelpG charges (Gaussian09 at the HF/6-31G* level), followed by two-stage RESP fitting. The absolute hydration free energy was computed by an established protocol including alchemical free energy perturbation with thermodynamic integration. The use of standard OPLS-AA charges (protocol 1) with a number of newly parametrized charges and the use of histidine derived parameters for imidazole yielded an overall root mean square deviation of the prediction from the experimental data of 1.75 kcal/mol. The precision of our results appears to be mainly limited by relatively poor reproducibility of the Lennard-Jones contribution towards the solvation free energy, for which we observed large variability that could be traced to a strong dependence on the initial system conditions.     

Extended solvent-contact model approach to SAMPL4 blind prediction challenge for hydration free energies

Extended solvent-contact model was applied to the blind prediction of the hydration free energies of 47 organic molecules included in the SAMPL4 data set. To obtain a suitable prediction tool, we constructed a hydration free energy function involving three kinds of atomic parameters. With respect to total 34 atom types introduced to describe all SAMPL4 molecules, 102 atomic parameters were defined and optimized with a standard genetic algorithm in such a way to minimize the difference between the experimental hydration free energies and those calculated with the hydration free energy function. In this parameterization, we used a training set comprising 77 organic molecules with varying sizes and shapes. The estimated hydration free energies for the SAMPL4 molecules compared reasonably well with the experimental results with the associated squared correlation coefficient and root mean square deviation of 0.89 and 1.46 kcal/mol, respectively. Based on the comparative analysis of experimental and computational hydration free energies of the SAMPL4 molecules, the methods for further improvement of the present hydration model are suggested.     

Fluctuating charge normal modes: An algorithm for implementing molecular dynamics simulations with polarizable potentials

A new method for performing molecular dynamics simulations with fluctuating charge polarizable potentials is introduced. In fluctuating charge models, polarizability is treated by allowing the partial charges to be variables, with values that are coupled to charges on the same molecule as well as those on other molecules. The charges can be efficiently propagated in a molecular dynamics simulation using extended Lagrangian dynamics. By making a coordinate change from the charge variables to a set of normal mode charge coordinates for each molecule, a new method is constructed in which the normal mode charge variables uncouple from those on the same molecule. The method is applied to the TIP4P-FQ model of water and compared to other methods for implementing the dynamics. The methods are compared using different molecular dynamics time steps.     

Predictions of hydration free energies from continuum solvent with solute polarizable models: the SAMPL2 blind challenge

This paper reports the results of our attempt to predict hydration free energies on the SAMPL2 blind challenge dataset. We mostly examine the effects of the solute electrostatic component on the accuracy of the predictions. The usefulness of electronic polarization in predicting hydration free energies is assessed by comparing the Electronic Polarization from Internal Continuum model and the self consistent reaction field IEF-PCM to standard non-polarizable charge models such as RESP and AM1-BCC. We also determine an optimal restraint weight for Dielectric-RESP atomic charges fitting. Statistical analysis of the results could not distinguish the methods from one another. The smallest average unsigned error obtained is 1.9 ± 0.6 kcal/mol (95% confidence level). A class of outliers led us to investigate the importance of the solute–solvent instantaneous induction energy, a missing term in PB continuum models. We estimated values between −1.5 and −6 kcal/mol for a series of halo-benzenes which can explain why some predicted hydration energies of non-polar molecules significantly disagreed with experiment.     

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.     

Molecular dynamics simulations of peptide carboxylate hydration

Aqueous solvation of carboxylate groups, as present in the glycine zwitterion and the dipeptide aspartylalanine, is studied employing a force-field that includes distributed multipole electrostatics and induction contributions (Amoebapro: P. Ren and J. W. Ponder, J. Comput. Chem., 2002, 23, 1497; P. Ren and J. W. Ponder, J. Phys. Chem. B, 2003, 107, 5933; J. W. Ponder and D. A. Case, Adv. Protein Chem., 2003, 66, 27). Radial and orientation distribution functions, as well as hydration numbers, are calculated and compared with existing simulation data derived from Car-Parrinello molecular dynamics (CPMD), and also distributed-charge force-fields. Connections are also made with experimental data for solvation of carboxylates in water. Our findings show that Amoebapro yields carboxylate solvation properties in very good agreement with CPMD results, significantly closer agreement than can be obtained from traditional force-fields. We also demonstrate that the influence of solvation on the conformation of the dipeptide is markedly different using Amoebapro compared with the other force-fields.     

Prediction of free energies of hydration with COSMO-RS on the SAMPL4 data set

The COSMO-RS method has been used for the prediction of free energies of hydration on a dataset of 47 complex multifunctional compounds considered in the SAMPL4 challenge. Straight application of the COSMOtherm software with the parameterization C21_0108 yields a predictive accuracy of 1.46 kcal/mol root mean square error overall and 1.18 kcal/mol if a single dominant outlier is removed.     

Prediction of cyclohexane-water distribution coefficients for the SAMPL5 data set using molecular dynamics simulations with the OPLS-AA force field

All-atom molecular dynamics simulations were used to predict water-cyclohexane distribution coefficients \(D_{cw}\) of a range of small molecules as part of the SAMPL5 blind prediction challenge. Molecules were parameterized with the transferable all-atom OPLS-AA force field, which required the derivation of new parameters for sulfamides and heterocycles and validation of cyclohexane parameters as a solvent. The distribution coefficient was calculated from the solvation free energies of the compound in water and cyclohexane. Absolute solvation free energies were computed by an established protocol using windowed alchemical free energy perturbation with thermodynamic integration. This protocol resulted in an overall root mean square error in \(\log D_{cw}\) of almost 4 log units and an overall signed error of ?3 compared to experimental data. There was no substantial overall difference in accuracy between simulating in NVT and NPT ensembles. The signed error suggests a systematic error but the experimental \(D_{cw}\) data on their own are insufficient to uncover the source of this error. Preliminary work suggests that the major source of error lies in the hydration free energy calculations.     

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.     

Refinement of the primary hydration shell model for molecular dynamics simulations of large proteins

A realistic representation of water molecules is important in molecular dynamics simulation of proteins. However, the standard method of solvating biomolecules, that is, immersing them in a box of water with periodic boundary conditions, is computationally expensive. The primary hydration shell (PHS) method, developed more than a decade ago and implemented in CHARMM, uses only a thin shell of water around the system of interest, and so greatly reduces the computational cost of simulations. Applying the PHS method, especially to larger proteins, revealed that further optimization and a partial reworking was required and here we present several improvements to its performance. The model is applied to systems with different sizes, and both water and protein behaviors are compared with those observed in standard simulations with periodic boundary conditions and, in some cases, with experimental data. The advantages of the modified PHS method over its original implementation are clearly apparent when it is applied to simulating the 82 kDa protein Malate Synthase G. © 2009 Wiley Periodicals, Inc. J Comput Chem 2009     

Predicting hydration free energies with a hybrid QM/MM approach: an evaluation of implicit and explicit solvation models in SAMPL4

The correct representation of solute-water interactions is essential for the accurate simulation of most biological phenomena. Several highly accurate quantum methods are available to deal with solvation by using both implicit and explicit solvents. So far, however, most evaluations of those methods were based on a single conformation, which neglects solute entropy. Here, we present the first test of a novel approach to determine hydration free energies that uses molecular mechanics (MM) to sample phase space and quantum mechanics (QM) to evaluate the potential energies. Free energies are determined by using re-weighting with the Non-Boltzmann Bennett (NBB) method. In this context, the method is referred to as QM-NBB. Based on snapshots from MM sampling and accounting for their correct Boltzmann weight, it is possible to obtain hydration free energies that incorporate the effect of solute entropy. We evaluate the performance of several QM implicit solvent models, as well as explicit solvent QM/MM for the blind subset of the SAMPL4 hydration free energy challenge. While classical free energy simulations with molecular dynamics give root mean square deviations (RMSD) of 2.8 and 2.3 kcal/mol, the hybrid approach yields an improved RMSD of 1.6 kcal/mol. By selecting an appropriate functional and basis set, the RMSD can be reduced to 1 kcal/mol for calculations based on a single conformation. Results for a selected set of challenging molecules imply that this RMSD can be further reduced by using NBB to reweight MM trajectories with the SMD implicit solvent model.     

The SAMPL2 blind prediction challenge: introduction and overview

The interactions between a molecule and the aqueous environment underpin any process that occurs in solution, from simple chemical reactions to protein–ligand binding to protein aggregation. Fundamental measures of the interaction between molecule and aqueous phase, such as the transfer energy between gas phase and water or the energetic difference between two tautomers of a molecule in solution, remain nontrivial to predict accurately using current computational methods. SAMPL2 represents the third annual blind prediction of transfer energies, and the first time tautomer ratios were included in the challenge. Over 60 sets of predictions were submitted, and each participant also attempted to estimate the error in their predictions, a task that proved difficult for most. The results of this blind assessment of the state of the field for transfer energy and tautomer ratio prediction both indicate where the field is performing well and point out flaws in current methods.     

Blind prediction of solvation free energies from the SAMPL4 challenge

Here, we give an overview of the small molecule hydration portion of the SAMPL4 challenge, which focused on predicting hydration free energies for a series of 47 small molecules. These gas-to-water transfer free energies have in the past proven a valuable test of a variety of computational methods and force fields. Here, in contrast to some previous SAMPL challenges, we find a relatively wide range of methods perform quite well on this test set, with RMS errors in the 1.2 kcal/mol range for several of the best performing methods. Top-performers included a quantum mechanical approach with continuum solvent models and functional group corrections, alchemical molecular dynamics simulations with a classical all-atom force field, and a single-conformation Poisson–Boltzmann approach. While 1.2 kcal/mol is still a significant error, experimental hydration free energies covered a range of nearly 20 kcal/mol, so methods typically showed substantial predictive power. Here, a substantial new focus was on evaluation of error estimates, as predicting when a computational prediction is reliable versus unreliable has considerable practical value. We found, however, that in many cases errors are substantially underestimated, and that typically little effort has been invested in estimating likely error. We believe this is an important area for further research.     

vmdICE: a plug-in for rapid evaluation of molecular dynamics simulations using VMD

Molecular dynamics (MD) is a powerful in silico method to investigate the interactions between biomolecules. It solves Newton's equations of motion for atoms over a specified period of time and yields a trajectory file, containing the different spatial arrangements of atoms during the simulation. The movements and energies of each single atom are recorded. For evaluating of these simulation trajectories with regard to biomedical implications, several methods are available. Three well-known ones are the root mean square deviation (RMSD), the root mean square fluctuation (RMSF) and solvent accessible surface area (SASA). Herein, we present a novel plug-in for the software "visual molecular dynamics" (VMD) that allows an interactive 3D representation of RMSD, RMSF, and SASA, directly on the molecule. On the one hand, our plug-in is easy to handle for inexperienced users, and on the other hand, it provides a fast and flexible graphical impression of the spatial dynamics of a system for experts in the field.     

The SAMPL3 blind prediction challenge: transfer energy overview

Prediction of the free energy of solvation of a small molecule, or its transfer energy, is a necessary step along the path towards calculating the interactions between molecules that occur in an aqueous environment. A set of these transfer energies were gathered from the literature for series of chlorinated molecules with varying numbers of chlorines based on ethane, biphenyl, and dibenzo-p-dioxin. This focused set of molecules were then provided as a blinded challenge to assess the ability of current computational solvation methods to accurately model the interactions between water and increasingly chlorinated compounds. This was presented as part of the SAMPL3 challenge, which represented the fourth iterative blind prediction challenge involving transfer energies. The results of this exercise demonstrate that the field in general has difficulty predicting the transfer energies of more highly chlorinated compounds, and that methods seem to be erring in the same direction.     

Ab initio molecular dynamics simulations of the Li4F4 cluster

Molecular dynamics simulations have been performed directly on the ab initio potential energy surface of Li₄F₄, which was generated within the Hartree-Fock approximation using a Gaussian basis set (split valence contraction). Trajectories at different temperatures yield the temperature dependence of the infrared spectra and the photoelectron spectra. For the infrared spectra comparison is made with MD results using a shell model ion pair potential function.     

Blind prediction of HIV integrase binding from the SAMPL4 challenge

Here, we give an overview of the protein-ligand binding portion of the Statistical Assessment of Modeling of Proteins and Ligands 4 (SAMPL4) challenge, which focused on predicting binding of HIV integrase inhibitors in the catalytic core domain. The challenge encompassed three components—a small “virtual screening” challenge, a binding mode prediction component, and a small affinity prediction component. Here, we give summary results and statistics concerning the performance of all submissions at each of these challenges. Virtual screening was particularly challenging here in part because, in contrast to more typical virtual screening test sets, the inactive compounds were tested because they were thought to be likely binders, so only the very top predictions performed significantly better than random. Pose prediction was also quite challenging, in part because inhibitors in the set bind to three different sites, so even identifying the correct binding site was challenging. Still, the best methods managed low root mean squared deviation predictions in many cases. Here, we give an overview of results, highlight some features of methods which worked particularly well, and refer the interested reader to papers in this issue which describe specific submissions for additional details.