Generalized Born Implicit Solvent Research Articles

Tuning Potential Functions to Host-Guest Binding Data.

Software to more rapidly and accurately predict protein-ligand binding affinities is of high interest for early-stage drug discovery, and physics-based methods are among the most widely used technologies for this purpose. The accuracy of these methods depends critically on the accuracy of the potential functions that they use. Potential functions are typically trained against a combination of quantum chemical and experimental data. However, although binding affinities are among the most important quantities to predict, experimental binding affinities have not to date been integrated into the experimental data set used to train potential functions. In recent years, the use of host-guest complexes as simple and tractable models of binding thermodynamics has gained popularity due to their small size and simplicity, relative to protein-ligand systems. Host-guest complexes can also avoid ambiguities that arise in protein-ligand systems such as uncertain protonation states. Thus, experimental host-guest binding data are an appealing additional data type to integrate into the experimental data set used to optimize potential functions. Here, we report the extension of the Open Force Field Evaluator framework to enable the systematic calculation of host-guest binding free energies and their gradients with respect to force field parameters, coupled with the curation of 126 host-guest complexes with available experimental binding free energies. As an initial application of this novel infrastructure, we optimized generalized Born (GB) cavity radii for the OBC2 GB implicit solvent model against experimental data for 36 host-guest systems. This refitting led to a dramatic improvement in accuracy for both the training set and a separate test set with 90 additional host-guest systems. The optimized radii also showed encouraging transferability from host-guest systems to 59 protein-ligand systems. However, the new radii are significantly smaller than the baseline radii and lead to excessively favorable hydration free energies (HFEs). Thus, users of the OBC2 GB model currently may choose between GB cavity radii that yield more accurate binding affinities and GB cavity radii that yield more accurate HFEs. We suspect that achieving good accuracy on both will require more far-reaching adjustments to the GB model. We note that binding free-energy calculations using the OBC2 model in OpenMM gain about a 10× speedup relative to corresponding explicit solvent calculations, suggesting a future role for implicit solvent absolute binding free-energy (ABFE) calculations in virtual compound screening. This study proves the principle of using host-guest systems to train potential functions that are transferrable to protein-ligand systems and provides an infrastructure that enables a range of applications.

A method for protein structure sampling under three-dimensional reference interaction site model solvation based on hybrid Monte Carlo framework

An efficient sampling method for protein structure using molecular dynamics (MD) simulation and a three-dimensional reference interaction-site model (3D-RISM) theory has been developed based on the hybrid Monte Carlo framework. In this method, new structures are generated by MD simulation with the generalized Born implicit solvent model and the Metropolis-like criterion to accept or reject the structure candidate is applied based on the free energy corresponding to the 3D-RISM Hamiltonian at each specified time propagation interval. The resulting ensemble of structures satisfies the 3D-RISM Hamiltonian-based equilibrium distribution. The method was applied to alanine dipeptide and methionine-enkephalin polypeptides in aqueous solution as test cases. The results demonstrated the efficiency of the method.

Replica exchange molecular dynamics simulations reveal self-association sites in M-crystallin caused by mutations provide insights of cataract

Crystallins are ubiquitous, however, prevalence is seen in eye lens. Eye lens crystallins are long-lived and structural intactness is required for maintaining lens transparency and protein solubility. Mutations in crystallins often lead to cataract. In this study, we performed mutations at specific sites of M-crystallin, a close homologue of eye lens crystallin and studied by using replica exchange molecular dynamics simulation with generalized Born implicit solvent model. Mutations were made on the Ca2+ binding residues (K34D and S77D) and in the hydrophobic core (W45R) which is known to cause congenital cataract in homologous γD-crystallin. The chosen mutations caused large motion of the N-terminal Greek key, concomitantly broke the interlocking Greek keys interactions and perturbed the compact core resulting in several folded and partially unfolded states. Partially unfolded states exposed large hydrophobic patches that could act as precursors for self-aggregation. Accumulation of such aggregates is the potential cause of cataract in homologous eye lens crystallins.

Multidimensional Global Optimization and Robustness Analysis in the Context of Protein-Ligand Binding.

Accuracy of protein-ligand binding free energy calculations utilizing implicit solvent models is critically affected by parameters of the underlying dielectric boundary, specifically, the atomic and water probe radii. Here, a global multidimensional optimization pipeline is developed to find optimal atomic radii specifically for protein-ligand binding calculations in implicit solvent. The computational pipeline has these three key components: (1) a massively parallel implementation of a deterministic global optimization algorithm (VTDIRECT95), (2) an accurate yet reasonably fast generalized Born implicit solvent model (GBNSR6), and (3) a novel robustness metric that helps distinguish between nearly degenerate local minima via a postprocessing step of the optimization. A graph-based "kT-connectivity" approach to explore and visualize the multidimensional energy landscape is proposed: local minima that can be reached from the global minimum without exceeding a given energy threshold (kT) are considered to be connected. As an illustration of the capabilities of the optimization pipeline, we apply it to find a global optimum in the space of just five radii: four atomic (O, H, N, and C) radii and water probe radius. The optimized radii, ρW = 1.37 Å, ρC = 1.40 Å, ρH = 1.55 Å, ρN = 2.35 Å, and ρO = 1.28 Å, lead to a closer agreement of electrostatic binding free energies with the explicit solvent reference than two commonly used sets of radii previously optimized for small molecules. At the same time, the ability of the optimizer to find the global optimum reveals fundamental limits of the common two-dielectric implicit solvation model: the computed electrostatic binding free energies are still almost 4 kcal/mol away from the explicit solvent reference. The proposed computational approach opens the possibility to further improve the accuracy of practical computational protocols for binding free energy calculations.

Mechanistic Insights from Replica Exchange Molecular Dynamics Simulations into Mutation Induced Disordered-to-Ordered Transition in Hahellin, a βγ-Crystallin.

Intrinsically disordered proteins (IDPs) form a special category because they lack a unique well-folded 3D structure under physiological conditions. They play crucial role in cell signaling and regulatory functions and are responsible for several diseases. Although they are abundant in nature, only a small fraction of them have been characterized until date. Such proteins adopt a range of conformations and can undergo transformation from disordered-to-ordered state or vice versa upon binding to ligand. Insights of such conformational transition is perplexing in several cases. In the present study, we characterized disordered as well as ordered states and the interactions contributing the transitions through a mutational study by employing replica exchange molecular dynamics simulation with generalized Born implicit solvent model on a protein from the βγ-crystallin superfamily. Most of the proteins within this superfamily are inherently ordered and highly stable. However, Hahellin, although a member of the βγ-crystallin family, is intrinsically disordered in its apo-form which takes a well-ordered βγ-crystallin fold upon binding to Ca2+. It is intriguing that the mutation at the fifth position of the canonical motif to Arg increases the domain stability in several ordered microbial βγ-crystallins with concomitant loss in Ca2+ binding affinity. We carried out similar Ser to Arg mutations at fifth position of the canonical motif for the first time in an intrinsically disordered protein to understand the mechanistic insights of conformational transition. Our study revealed that newly formed ionic and hydrogen bonding interactions at the canonical Ca2+ binding sites play a crucial role in transforming the disordered conformation into ordered βγ-crystallin.

Unraveling the structural and molecular properties of 34-residue levans with various branching degrees by replica exchange molecular dynamics simulations.

Levan has various potential applications in the pharmaceutical and food industries, such as cholesterol-lowering agents and prebiotics, due to its beneficial properties, which depend on its length and branching degree. A previous study also found that the branching degree of levan affected anti-tumor activities against SNU-1 and HepG2 tumor cell lines. Despite its promising potential, the properties of levans with different branching degrees are not well understood at the molecular level. In two models of the generalized Born implicit solvent (GBHCT and GBOBC1), we employed replica-exchange molecular dynamics simulations to explore conformational spaces of 34-residue levans (L34) with branching degrees of zero (LFO34B0), one (LFO34B1), three (LFO34B3) and five (LFO34B5), as well as to elucidate their structural and molecular properties. To ensure a fair comparison of the effects of branching degree on these properties, we focused on analyzing the properties of the central 21-residue of the main chains of all systems. Our results show that all major representative conformations tend to form helix-like structures with kinks, where two-kink helix-like structures have the highest population. As branching degree increases, the population of helix-like structures with zero or one kink tends to increase slightly. As the number of kinks in the structures with the same branching degree increases, the average values of the lengths and angles among centers of masses of three consecutive turns of residue i, i+3, and i+6 tended to decrease. Due to its highest occurring frequencies, the O6 (i)—H3O (i+1) hydrogen bond could be important for helix-like structure formation. Moreover, hydrogen bonds forming among the branching residue (br), branching position (bp) and other residues of L34B1, L34B3 and L34B5 were identified. The O1(bp)—H3O(br), O1(br)—H3O(br) and O5(br)—H1O(br) hydrogen bonds were found in the first-, second- and third-highest occurrence frequencies, respectively. Our study provides novel and important insights into conformational spaces and the structural and molecular properties of 34-residue levans with various branching degrees, which tend to form helix-like structures with kinks.

Generalized Born implicit solvent models for small molecule hydration free energies.

Hydration free energy estimation of small molecules from all-atom simulations was widely investigated in recent years, as it provides an essential test of molecular force fields and our understanding of solvation effects. While explicit solvent representations result in highly accurate models, they also require extensive sampling due to the high number of solvent degrees of freedom. Implicit solvent models, such as those based on the generalized Born model for electrostatic solvation effects and a solvent accessible surface area term for nonpolar contributions (GBSA), significantly reduce the number of degrees of freedom and the computational cost to estimate hydration free energies. However, a recent survey revealed a gap in the accuracy between explicit TIP3P solvent estimates and those computed with many common GBSA models. Here we address this shortcoming by providing a thorough comparison of the performance of three implicit solvent models with different nonpolar contributions and a generalized Born term to estimate experimental hydration free energies. Starting with a minimal set of only ten atom types, we demonstrate that a nonpolar term with atom type dependent surface tension coefficients in combination with an accurate generalized Born term and fully optimized parameters performs best in estimating hydration free energies, even yielding comparable results to the explicit TIP3P water model. Analysis of our results provides evidence that the asymmetric behavior of water around oppositely charged atoms is one of the main sources of error for two of the three implicit solvent models. Explicitly accounting for this effect in the parameterization reduces the corresponding errors, suggesting this as a general strategy for improving implicit solvent models. The findings presented here will help to improve the existing generalized Born based implicit solvent models implemented in state-of-the-art molecular simulation packages.

Replica exchange molecular dynamics simulations reveal the structural and molecular properties of levan-type fructo-oligosaccharides of various chain lengths

BackgroundLevan and levan-type fructo-oligosaccharides (LFOs) have various potential applications in pharmaceutical and food industries due to their beneficial properties such as their low intrinsic viscosity and high water solubility. Previous studies showed that they exhibited prebiotic effects, anti-inflammatory and anti-tumor activities against Sarcoma-180 tumor cells of human. Despite their various potential applications, the structural and molecular properties of LFOs of various chain lengths are not well understood.ResultsWe employed the replica-exchange molecular dynamics simulations method (REMD) in AMBER14 to elucidate structural and molecular properties of LFOs with chain lengths of 5 (LFO5), 10 (LFO10) and 15 (LFO15) residues in two models of generalized Born implicit solvent (GBHCT and GBOBC1). For LFO10 and LFO15, four distinct conformations (helix-like, partial helix, zig-zag and random structures) were characterized by their upper-middle and lower-middle torsions. For LFO5, two distinct conformations (partial helix and random structures) were characterized by their middle torsion and molecular angle of residues 1, 3 and 5. To determine hydrogen bonds important for the formation of helix-like structures of LFO10 and LFO15, occurrence frequencies of hydrogen bonds were analyzed, and the O6(i)--H3O(i+1) hydrogen bond was found with the highest frequency, suggesting its importance in helix formation. Among three dihedral angles between two fructosyl units [ϕ (O5’-C2’-O6-C6), ψ (C2’-O6-C6-C5) and ω (O6-C6-C5-C4)], dihedral angle distributions showed that ω was the most flexible dihedral angle and probably responsible for conformational differences of LFOs.ConclusionsOur study provides important insights into the structural and molecular properties of LFOs, which tend to form helical structures as the chain length increases from 5 to 15 residues. This information could be beneficial for the selection of LFOs with appropriate lengths and properties for pharmaceutical and biological applications.Electronic supplementary materialThe online version of this article (doi:10.1186/s12859-016-1182-7) contains supplementary material, which is available to authorized users.

Protein-Ligand Electrostatic Binding Free Energies from Explicit and Implicit Solvation.

Accurate yet efficient computational models of solvent environment are central for most calculations that rely on atomistic modeling, such as prediction of protein-ligand binding affinities. In this study, we evaluate the accuracy of a recently developed generalized Born implicit solvent model, GBNSR6 (Aguilar et al. J. Chem. Theory Comput. 2010, 6, 3613-3639), in estimating the electrostatic solvation free energies (ΔG(pol)) and binding free energies (ΔΔG(pol)) for small protein-ligand complexes. We also compare estimates based on three different explicit solvent models (TIP3P, TIP4PEw, and OPC). The two main findings are as follows. First, the deviation (RMSD = 7.04 kcal/mol) of GBNSR6 binding affinities from commonly used TIP3P reference values is comparable to the deviations between explicit models themselves, e.g. TIP4PEw vs TIP3P (RMSD = 5.30 kcal/mol). A simple uniform adjustment of the atomic radii by a single scaling factor reduces the RMS deviation of GBNSR6 from TIP3P to within the above "error margin" - differences between ΔΔG(pol) estimated by different common explicit solvent models. The simple radii scaling virtually eliminates the systematic deviation (ΔΔG(pol)) between GBNSR6 and two out of the three explicit water models and significantly reduces the deviation from the third explicit model. Second, the differences between electrostatic binding energy estimates from different explicit models is disturbingly large; for example, the deviation between TIP4PEw and TIP3P estimates of ΔΔG(pol) values can be up to ∼50% or ∼9 kcal/mol, which is significantly larger than the "chemical accuracy" goal of ∼1 kcal/mol. The absolute ΔG(pol) calculated with different explicit models could differ by tens of kcal/mol. These discrepancies point to unacceptably high sensitivity of binding affinity estimates to the choice of common explicit water models. The absence of a clear "gold standard" among these models strengthens the case for the use of accurate implicit solvation models for binding energetics, which may be orders of magnitude faster.

Comparison of NMR structures refined under implicit and explicit solvents

Refinements with atomistic molecular dynamics (MD) simulation have contributed to improving the qualities of NMR structures. In most cases, the calculations with atomistic MD simulation for NMR structures employ generalized-Born implicit solvent model (GBIS) to take into accounts solvation effects. Developments in algorithms and computational capacities have ameliorated GBIS to approximate solvation effects that explicit solvents bring about. However, the quantitative comparison of NMR structures in the latest GBIS and explicit solvents is lacking. In this study, we report the direct comparison of NMR structures that atomistic MD simulation coupled with GBIS and water molecules refined. Two model proteins, GB1 and ubiquitin, were recalculated with experimental distance and torsion angle restraints, under a series of simulated annealing time steps. Whereas the root mean square deviations of the resulting structures were apparently similar, AMBER energies, the most favored regions in Ramachandran plot, and MolProbity clash scores witnessed that GBIS-refined structures had the better geometries. The outperformance by GBIS was distinct in the structure calculations with sparse experimental restraints. We show that the superiority stemmed, at least in parts, from the inclusion of all the pairs of non-bonded interactions. The shorter computational times with GBIS than those for explicit solvents makes GBIS a powerful method for improving structural qualities particularly under the conditions that experimental restraints are insufficient. We also propose a method to separate the native-like folds from non-violating diverged structures.

Rapid Alchemical Free Energy Calculation Employing a Generalized Born Implicit Solvent Model

Generalized Born (GB) implicit solvent models are typically used in postprocessing of molecular dynamics trajectories obtained from explicit solvent simulations to estimate binding free energies or effects of mutations in proteins. The possibility to employ a GB implicit solvent model for the calculation of rigorous free energy changes associated with alchemical transformations has been explored. During free energy perturbation (FEP) simulations, Lennard-Jones, Coulomb, and Born radii parameters are transformed gradually in a single-topology series. The FEP calculations are embedded in a replica exchange scheme allowing rapid convergence. The method was tested on the calculation of relative hydration free energies, relative binding free energies of a ligand-receptor system, and in silico alanine scanning of a peptide-protein complex. In all cases, good agreement with available experimental data was obtained. On medium sized protein-ligand systems and using a cluster of graphical processing units, the approach allows the calculation of relative free energy changes associated with a chemical modification of a binding partner within a few minutes of computer time and opens the possibility for systematic in silico studies.

Systematic Assessment of the Effects of an All-Atom Force Field and the Implicit Solvent Model on the Refinement of NMR Structures with Subsets of Distance Restraints

Jun-Goo JeeResearch Institute of Pharmaceutical Sciences, College of Pharmacy, Kyungpook National University, Daegu 702-701, Korea. E-mail: jjee@knu.ac.krReceived February 22, 2014, Accepted March 5, 2014Employment of a time consuming, sophisticated calculation using the all-atom force field and generalized-Born implicit solvent model (GBIS) for refinement of NMR structures has become practical through advancesin computational methods and capacities. GBIS refinement improves the qualities of the resulting NMRstructures with reduced computational times. However, the contribution of GBIS to NMR structures has notbeen sufficiently studied in a quantitative way. In this paper, we report the effects of GBIS on the refined NMRstructures of ubiquitin (UBQ) and GB1 with subsets of distance restraints derived from experimental data.Random omission prepared a series of distance restraints 0.05, 0.1, 0.3, 0.5, and 0.7 times smaller. For eachnumber, we produced five different restraints for statistical analysis. We then recalculated the NMR structuresusing CYANA software, followed by GBIS refinements using the AMBER package. GBIS improved both theprecision and accuracy of all the structures, but to varied levels. The degrees of improvement were significantwhen the input restraints were insufficient. In particular, GBIS enabled GB1 to form an accurate structure evenwith distance restraints of 5%, revealing that the root-mean-square deviation was less than 1 A from the X-raybackbone structure. We also showed that the efficiency of searching the conformational space was moreimportant for finding accurate structures with the calculation of UBQ with 5% distance restraints than thenumber of conformations generated. Our data will provide a meaningful guideline to judge and compare thestructural improvements by GBIS. Key Words : NMR, Generalized-Born implicit solvent model, Ubiquitin, GB1IntroductionSteady advances in algorithms for both NOE assignmentsand structure calculations have made automatic calculationsof protein 3D structures with raw NOE data, provided thechemical shifts are assigned for most atoms and sufficientpeaks exist.

Evaluation of Generalized Born Model Accuracy for Absolute Binding Free Energy Calculations.

Generalized Born (GB) implicit solvent models are widely used in molecular dynamics simulations to evaluate the interactions of biomolecular complexes. The continuum treatment of the solvent results in significant computational savings in comparison to an explicit solvent representation. It is, however, not clear how accurately the GB approach reproduces the absolute free energies of biomolecular binding. On the basis of induced dissociation by means of umbrella sampling simulations, the absolute binding free energies of small proline-rich peptide ligands and a protein receptor were calculated. Comparative simulations according to the same protocol were performed by employing an explicit solvent model and various GB-type implicit solvent models in combination with a nonpolar surface tension term. The peptide ligands differed in a key residue at the peptide-protein interface, including either a nonpolar, a neutral polar, a positively charged, or a negatively charged group. For the peptides with a neutral polar or nonpolar interface residue, very good agreement between the explicit solvent and GB implicit solvent results was found. Deviations in the main separation free energy contributions are smaller than 1 kcal/mol. In contrast, for peptides with a charged interface residue, significant deviations of 2-4 kcal/mol were observed. The results indicate that recent GB models can compete with explicit solvent representations in total binding free energy calculations as long as no charged residues are present at the binding interface.

Effects of force fields for refining protein NMR structures with atomistic force fields and generalized-Born implicit solvent model

Atomistic molecular dynamics (MD) simulation has become mature enabling close approximation of the real behaviors of biomolecules. In biomolecular NMR field, atomistic MD simulation coupled with generalized implicit solvent model (GBIS) has contributed to improving the qualities of NMR structures in the refinement stage with experimental restraints. Here all-atom force fields play important roles in defining the optimal positions between atoms and angles, resulting in more precise and accurate structures. Despite successful applications in refining NMR structure, however, the research that has studied the influence of force fields in GBIS is limited. In this study, we compared the qualities of NMR structures of two model proteins, ubiquitin and GB1, under a series of AMBER force fields-ff99SB, ff99SB-ILDN, ff99SB-NMR, ff12SB, and ff13-with experimental restraints. The root mean square deviations of backbone atoms and packing scores that reflect the apparent structural qualities were almost indistinguishable except ff13. Qualitative comparison of parameters, however, indicates that ff99SB-ILDN is more recommendable, at least in the cases of ubiquitin and GB1.

Constant pH Replica Exchange Molecular Dynamics in Explicit Solvent Using Discrete Protonation States: Implementation, Testing, and Validation.

By utilizing Graphics Processing Units, we show that constant pH molecular dynamics simulations (CpHMD) run in Generalized Born (GB) implicit solvent for long time scales can yield poor pKa predictions as a result of sampling unrealistic conformations. To address this shortcoming, we present a method for performing constant pH molecular dynamics simulations (CpHMD) in explicit solvent using a discrete protonation state model. The method involves standard molecular dynamics (MD) being propagated in explicit solvent followed by protonation state changes being attempted in GB implicit solvent at fixed intervals. Replica exchange along the pH-dimension (pH-REMD) helps to obtain acceptable titration behavior with the proposed method. We analyzed the effects of various parameters and settings on the titration behavior of CpHMD and pH-REMD in explicit solvent, including the size of the simulation unit cell and the length of the relaxation dynamics following protonation state changes. We tested the method with the amino acid model compounds, a small pentapeptide with two titratable sites, and hen egg white lysozyme (HEWL). The proposed method yields superior predicted pKa values for HEWL over hundreds of nanoseconds of simulation relative to corresponding predicted values from simulations run in implicit solvent.

Molecular dynamics study of DNA oligomers under angled pulling

Extensive 13-microsecond steered molecular dynamics simulations (SMD) were performed to study the mechanical properties of short double-stranded DNA under tensile loads at the atomistic level. We elucidate the influence of the pulling rate and the pulling angle on stretching behavior of the DNA duplex. Generalized Born implicit solvent methodology and Langevin dynamics were employed to represent the effects of aqueous solvation under physiological conditions. Our simulations show that for all pulling angles, within the range of pulling rates studied here, the DNA force-extension curves consist of three different regimes: elastic, strain softening and strain hardening phases. Angled pulling can, however, alter the mechanical properties unpredictably, especially in the overstretched regime. In addition, these properties may depend on the velocity at which the DNA is pulled making the problem even more complex. A detailed analysis of the base-pairing and base-stacking interactions during stretching is carried out to investigate the origin of the different DNA behaviors under various pulling angles and velocities.

Effects of generalized-Born implicit solvent models in NMR structure refinement

Rapid advances of computational power and method have made it practical to apply the time-consuming calculations with all-atom force fields and sophisticated potential energies into refining NMR structure. Added to the all-atom force field, generalized-Born implicit solvent model (GBIS) contributes substantially to improving the qualities of the resulting NMR structures. GBIS approximates the effects that explicit solvents bring about even with fairly reduced computational times. Although GBIS is employed in the final stage of NMR structure calculation with experimental restraints, the effects by GBIS on structures have been reported notable. However, the detailed effect is little studied in a quantitative way. In this study, we report GBIS refinements of ubiquitin and GB1 structures by six GBIS models of AMBER package with experimental distance and backbone torsion angle restraints. Of GBIS models tested, the calculations with igb

Structure refinement of protein model decoys requires accurate side-chain placement

In this study, the application of temperature-based replica-exchange (T-ReX) simulations for structure refinement of decoys taken from the I-TASSER dataset was examined. A set of eight nonredundant proteins was investigated using self-guided Langevin dynamics (SGLD) with a generalized Born implicit solvent model to sample conformational space. For two of the protein test cases, a comparison of the SGLD/T-ReX method with that of a hybrid explicit/implicit solvent molecular dynamics T-ReX simulation model is provided. Additionally, the effect of side-chain placement among the starting decoy structures, using alternative rotamer conformations taken from the SCWRL4 modeling program, was investigated. The simulation results showed that, despite having near-native backbone conformations among the starting decoys, the determinant of their refinement is side-chain packing to a level that satisfies a minimum threshold of native contacts to allow efficient excursions toward the downhill refinement regime on the energy landscape. By repacking using SCWRL4 and by applying the RWplus statistical potential for structure identification, the SGLD/T-ReX simulations achieved refinement to an average of 38% increase in the number of native contacts relative to the original I-TASSER decoy sets and a 25% reduction in values of C(α) root-mean-square deviation. The hybrid model succeeded in obtaining a sharper funnel to low-energy states for a modeled target than the implicit solvent SGLD model; yet, structure identification remained roughly the same. Without meeting a threshold of near-native packing of side chains, the T-ReX simulations degrade the accuracy of the decoys, and subsequently, refinement becomes tantamount to the protein folding problem.

SPFP: Speed without compromise—A mixed precision model for GPU accelerated molecular dynamics simulations

A new precision model is proposed for the acceleration of all-atom classical molecular dynamics (MD) simulations on graphics processing units (GPUs). This precision model replaces double precision arithmetic with fixed point integer arithmetic for the accumulation of force components as compared to a previously introduced model that uses mixed single/double precision arithmetic. This significantly boosts performance on modern GPU hardware without sacrificing numerical accuracy. We present an implementation for NVIDIA GPUs of both generalized Born implicit solvent simulations as well as explicit solvent simulations using the particle mesh Ewald (PME) algorithm for long-range electrostatics using this precision model. Tests demonstrate both the performance of this implementation as well as its numerical stability for constant energy and constant temperature biomolecular MD as compared to a double precision CPU implementation and double and mixed single/double precision GPU implementations.

Routine Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 1. Generalized Born

We present an implementation of generalized Born implicit solvent all-atom classical molecular dynamics (MD) within the AMBER program package that runs entirely on CUDA enabled NVIDIA graphics processing units (GPUs). We discuss the algorithms that are used to exploit the processing power of the GPUs and show the performance that can be achieved in comparison to simulations on conventional CPU clusters. The implementation supports three different precision models in which the contributions to the forces are calculated in single precision floating point arithmetic but accumulated in double precision (SPDP), or everything is computed in single precision (SPSP) or double precision (DPDP). In addition to performance, we have focused on understanding the implications of the different precision models on the outcome of implicit solvent MD simulations. We show results for a range of tests including the accuracy of single point force evaluations and energy conservation as well as structural properties pertainining to protein dynamics. The numerical noise due to rounding errors within the SPSP precision model is sufficiently large to lead to an accumulation of errors which can result in unphysical trajectories for long time scale simulations. We recommend the use of the mixed-precision SPDP model since the numerical results obtained are comparable with those of the full double precision DPDP model and the reference double precision CPU implementation but at significantly reduced computational cost. Our implementation provides performance for GB simulations on a single desktop that is on par with, and in some cases exceeds, that of traditional supercomputers.