Accelerating molecular dynamic simulation on graphics processing units

Friedrichs, Mark S.; Eastman, Peter; Vaidyanathan, Vishal; Houston, Michael J.; LeGrand, Scott; Beberg, Adam L.; Ensign, Daniel L.; Bruns, Christopher; Pande, Vijay S.

doi:10.1002/jcc.21209

Cited by 501 publications

(438 citation statements)

References 23 publications

Supporting

Mentioning

432

Contrasting

Unclassified

Order By: Relevance

“…Enormous progress has been made in all-atom calculations, due to extensive use of world-distributed computing (20), implementation of all-atom molecular dynamics software on graphical processor units (21), and, most notably, the construction of dedicated machines (22). However, coarsegrained approaches (23)(24)(25), in which several atoms are merged into single interaction sites, are used very extensively, because they enable us to treat protein systems at time and dimensional scales, which are orders of magnitude larger (26) than those possible in all-atom computations.…”

Section: Significancementioning

confidence: 99%

Lessons from application of the UNRES force field to predictions of structures of CASP10 targets

Mozolewska

Krupa

et al. 2013

Proc. Natl. Acad. Sci. U.S.A.

View full text Add to dashboard Cite

The performance of the physics-based protocol, whose main component is the United Residue (UNRES) physics-based coarse-grained force field, developed in our laboratory for the prediction of protein structure from amino acid sequence, is illustrated. Candidate models are selected, based on probabilities of the conformational families determined by multiplexed replica-exchange simulations, from the 10th Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction (CASP10). For target T0663, classified as a new fold, which consists of two α + β domains homologous to those of known proteins, UNRES predicted the correct symmetry of packing, in which the domains are rotated with respect to each other by 180°in the experimental structure. By contrast, models obtained by knowledge-based methods, in which each domain is modeled very accurately but not rotated, resulted in incorrect packing. Two UNRES models of this target were featured by the assessors. Correct domain packing was also predicted by UNRES for the homologous target T0644, which has a similar structure to that of T0663, except that the two domains are not rotated. Predictions for two other targets, T0668 and T0684_D2, are among the best ones by global distance test score. These results suggest that our physicsbased method has substantial predictive power. In particular, it has the ability to predict domain-domain orientations, which is a significant advance in the state of the art.protein folding | structure symmetry | multi-domain packing P rediction of protein structures from amino acid sequence still remains an unsolved problem of computational biology. Although, since the famous experiments by Anfinsen (1), it is known that a protein adopts the structure which is the (kinetically reachable) global minimum of the free energy of a system, it is not straightforward to implement this physical principle in practice because of the inaccuracy of existing force fields and because of the enormous difficulty to search the conformational space of the system. Therefore, the most effective methods for protein-structure prediction nowadays are knowledge-based approaches, in which database information is incorporated explicitly into the procedure (2). These methods can be divided into three categories, namely, comparative (homology) modeling (3-5), in which the target sequence is compared with the sequences for which experimental structures are known and those structures are usually selected as candidate models for which the greatest similarity is observed; threading (6-8), in which the target sequence is superposed on structures from a database, and those which give the highest score (lowest pseudoenergy) are selected as candidate predictions; and, finally, the fragment-assembly or minithreading method developed by David Baker and colleagues (9, 10), in which the predicted structure is assembled from nine-residue fragments extracted from a protein-structure database, and knowledge-and physicsbased filters are applied at each asse...

show abstract

Section: Significancementioning

confidence: 99%

Lessons from application of the UNRES force field to predictions of structures of CASP10 targets

Mozolewska

Krupa

et al. 2013

Proc. Natl. Acad. Sci. U.S.A.

View full text Add to dashboard Cite

show abstract

“…Today, novel computational technologies, such as graphical processor units (14)(15)(16)(17) and the specialized Anton computer (18), are enabling dramatically longer molecular dynamics (MD) simulations than hitherto feasible. The enormous conformational sampling power of these technologies has the potential to open a new window on the molecular mechanisms underlying the thermodynamics of biomolecular systems.…”

mentioning

confidence: 99%

Entropy–enthalpy transduction caused by conformational shifts can obscure the forces driving protein–ligand binding

Fenley

Muddana

Gilson

2012

Proc. Natl. Acad. Sci. U.S.A.

116

150

View full text Add to dashboard Cite

Molecular dynamics simulations of unprecedented duration now can provide new insights into biomolecular mechanisms. Analysis of a 1-ms molecular dynamics simulation of the small protein bovine pancreatic trypsin inhibitor reveals that its main conformations have different thermodynamic profiles and that perturbation of a single geometric variable, such as a torsion angle or interresidue distance, can select for occupancy of one or another conformational state. These results establish the basis for a mechanism that we term entropy-enthalpy transduction (EET), in which the thermodynamic character of a local perturbation, such as enthalpic binding of a small molecule, is camouflaged by the thermodynamics of a global conformational change induced by the perturbation, such as a switch into a high-entropy conformational state. It is noted that EET could occur in many systems, making measured entropies and enthalpies of folding and binding unreliable indicators of actual thermodynamic driving forces. The same mechanism might also account for the high experimental variance of measured enthalpies and entropies relative to free energies in some calorimetric studies. Finally, EET may be the physical mechanism underlying many cases of entropy-enthalpy compensation.T he entropic and enthalpic components of free energy can be informative regarding the mechanisms of biophysical processes, like protein folding and protein-ligand binding. Moreover, information on changes in entropy and enthalpy can usefully guide the design of improved drug molecules (1), with advantageous specificity (2) and physical properties (3). However, calorimetric studies of biomolecular binding and folding often reveal unexpected changes in entropy and enthalpy that are difficult to interpret in terms of physical driving forces (4-8). Some of these puzzling results are instances of entropy-enthalpy compensation (9), a common but not universal (10, 11) phenomenon in which perturbations of a system that increase the enthalpy also increase the entropy or vice versa; therefore, the net change in the free energy remains small. Experimental error in measured enthalpies, particularly when obtained by the relatively imprecise van 't Hoff method, can generate spurious entropy-enthalpy compensation (12); the nonphysical operation of Berkson's paradox (selection bias) (13) can also generate this compensation. However, analysis of collected calorimetric data for protein-ligand binding indicates that entropy-enthalpy compensation is a genuine and common physical phenomenon (13).Today, novel computational technologies, such as graphical processor units (14-17) and the specialized Anton computer (18), are enabling dramatically longer molecular dynamics (MD) simulations than hitherto feasible. The enormous conformational sampling power of these technologies has the potential to open a new window on the molecular mechanisms underlying the thermodynamics of biomolecular systems. For example, it can provide increasingly accurate estimates of thermodynamic quantities that ...

show abstract

“…A custom Python code making use of the GPUaccelerated OPENMM package [74][75][76] and the PYOPENMM Python wrapper 77 was used to conduct the simulations. Because OPENMM lacks a velocity Verlet integrator, a hybrid velocity Verlet 78 /leapfrog 79, 80 integration scheme was used for integration of the equations of motion, implemented as follows.…”

Section: Appendix A: Additional Simulation Details For Alanine Dipeptidementioning

confidence: 99%

Dynamical reweighting: Improved estimates of dynamical properties from simulations at multiple temperatures

Chodera

Swope²,

Noé³

et al. 2011

The Journal of Chemical Physics

Self Cite

View full text Add to dashboard Cite

Dynamical averages based on functionals of dynamical trajectories, such as time-correlation functions, play an important role in determining kinetic or transport properties of matter. At temperatures of interest, the expectations of these quantities are often dominated by contributions from rare events, making the precise calculation of these quantities by molecular dynamics simulation difficult. Here, we present a reweighting method for combining simulations from multiple temperatures (or from simulated or parallel tempering simulations) to compute an optimal estimate of the dynamical properties at the temperature of interest without the need to invoke an approximate kinetic model (such as the Arrhenius law). Continuous and differentiable estimates of these expectations at any temperature in the sampled range can also be computed, along with an assessment of the associated statistical uncertainty. For rare events, aggregating data from multiple temperatures can produce an estimate with the desired precision at greatly reduced computational cost compared with simulations conducted at a single temperature. Here, we describe use of the method for the canonical (NVT) ensemble using four common models of dynamics (canonical distribution of Hamiltonian trajectories, Andersen thermostatting, Langevin, and overdamped Langevin or Brownian dynamics), but it can be applied to any thermodynamic ensemble provided the ratio of path probabilities at different temperatures can be computed. To illustrate the method, we compute a time-correlation function for solvated terminallyblocked alanine peptide across a range of temperatures using trajectories harvested using a modified parallel tempering protocol.

show abstract

Accelerating molecular dynamic simulation on graphics processing units

Cited by 501 publications

References 23 publications

Lessons from application of the UNRES force field to predictions of structures of CASP10 targets

Lessons from application of the UNRES force field to predictions of structures of CASP10 targets

Entropy–enthalpy transduction caused by conformational shifts can obscure the forces driving protein–ligand binding

Dynamical reweighting: Improved estimates of dynamical properties from simulations at multiple temperatures

Contact Info

Product

Resources

About