Recent Advances in Simulation Software and Force Fields: Their Importance in Theoretical and Computational Chemistry and Biophysics
2024; American Chemical Society; Volume: 128; Issue: 49 Linguagem: Inglês
10.1021/acs.jpcb.4c06231
ISSN1520-6106
Autores Tópico(s)Enzyme Structure and Function
ResumoInfoMetricsFiguresRef. The Journal of Physical Chemistry BVol 128/Issue 49Article This publication is free to access through this site. Learn More CiteCitationCitation and abstractCitation and referencesMore citation options ShareShare onFacebookX (Twitter)WeChatLinkedInRedditEmailJump toExpandCollapse Special Issue PrefaceDecember 12, 2024Recent Advances in Simulation Software and Force Fields: Their Importance in Theoretical and Computational Chemistry and BiophysicsClick to copy article linkArticle link copied!Christophe Chipot*Christophe ChipotLaboratoire International Associé Centre National de la Recherche Scientifique et University of Illinois at Urbana−Champaign, UMR n°7019, Université de Lorraine, 54506 Vandoeuvre-lès-Nancy cedex, FranceDepartment of Physics and Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana−Champaign, Urbana, Illinois 61801, United StatesDepartment of Biochemistry and Molecular Biology, The University of Chicago, Chicago, Illinois 60637, United States*Email: [email protected]More by Christophe Chipothttps://orcid.org/0000-0002-9122-1698Open PDFThe Journal of Physical Chemistry BCite this: J. Phys. Chem. B 2024, 128, 49, 12023–12026Click to copy citationCitation copied!https://pubs.acs.org/doi/10.1021/acs.jpcb.4c06231https://doi.org/10.1021/acs.jpcb.4c06231Published December 12, 2024 Publication History Received 15 September 2024Published online 12 December 2024Published in issue 12 December 2024introductionCopyright © Published 2024 by American Chemical Society. This publication is available under these Terms of Use. Request reuse permissionsThis publication is licensed for personal use by The American Chemical Society. ACS PublicationsCopyright © Published 2024 by American Chemical SocietySubjectswhat are subjectsArticle subjects are automatically applied from the ACS Subject Taxonomy and describe the scientific concepts and themes of the article.Computational modelingComputer simulationsMolecular mechanicsSoftwareTheoretical and computational chemistrySpecial IssuePublished as part of The Journal of Physical Chemistry B special issue "Recent Advances in Simulation Software and Force Fields".The evolution of theoretical chemistry has been profoundly shaped by the development of sophisticated computational tools, which have revolutionized the way scientists tackle molecular systems. Structural biology and biophysics have particularly benefited from the emergence of equally apposite computer programs to address complex questions hitherto impossible to answer. Historically, these fields were primarily of an experimental nature, with theoretical insights often limited to rudimentary models and calculations. The landscape began to change in the middle of the last century with the advent of computers, which allowed complex calculations to be automated, marking the inception of computational chemistry as a discipline in its own right. In 1956, the first ab initio, Hartree–Fock calculations were conducted on diatomic molecules, (1) followed, the year after, by the first molecular dynamics (MD) simulation of elastic collisions between hard spheres. (2) The introduction of early software packages was a watershed moment, enabling the application of methods resting on the laws of not only classical mechanics, but also quantum mechanics to molecular systems that are far too complex to be handled analytically─and manually. As the computational power increased, so did the capabilities of computer codes, leading to the development of popular academic programs, most notably for MD simulations, as early as the 1980s. (3−5) These tools opened new frontiers, allowing the scientific community to simulate the dynamic behavior of macromolecules and explore processes like folding, (6) recognition and association, (7) and even catalysis (8) with unprecedented detail. The ability to model such dynamic processes over extended timescales transformed our understanding of molecular and structural biology, providing valuable insights that were previously accessible only through labor-intensive and time-consuming experimental approaches. The exponential growth in computational resources, combined with advances in software development and engineering, has since facilitated the investigation of increasingly large and intricate objects. Contemporary software packages that harness the ever increasing power of hundreds of computer nodes (9−13) are now capable of simulating entire organelles, viruses, and cellular environments, (14−17) modeling interactions between thousands of molecules simultaneously, while supplying detailed predictions of thermodynamic and kinetic properties. The integration of artificial intelligence into computational chemistry software represents the latest frontier, enabling the rapid screening of chemical spaces that would be infeasible to explore experimentally. (18,19) Machine-learning (ML) models, trained on vast data sets, have not only accelerated the pace of discovery, but they have also democratized the access to cutting-edge computational tools, empowering a community of researchers to push the boundaries of science. (20,21)While the software provides the computational framework for molecular simulations, the accuracy and reliability of these simulations hinge on the quality of the force fields employed. Force fields, which define the potential energy of a system as a function of atomic positions, have undergone significant refinement since their inception, evolving from simple models in the 1950s and 1960s to the highly sophisticated representations employed today. The earliest force fields were rudimentary, designed to describe the interactions within small organic molecules. (22) However, as the scope of computational chemistry expanded to include larger and more complex systems, most notably biomolecules, the limitations of these early models became apparent. (23) The development of biomolecular force fields in the 1980s and 1990s, like ECEPP, (24) AMBER, (25,26) CHARMM, (27) CVFF, (28) OPLS, (29) and GROMOS, (30) represented a significant leap forward. These force fields were parametrized using a combination of high-level quantum-mechanical (QM) calculations and experimental data, resulting in more accurate simulations of proteins, nucleic acids, and other complex biological systems. The refinement of these force fields has been a continuous and dynamic process, driven by an ever-growing body of experimental data and the insights gained from an interplay with computational studies. Modern force fields are highly parametrized, incorporating a very broad range of interactions, including electrostatic interactions that can go beyond the simplistic Coulombic, charge–charge term, (31) as well as an explicit account of more complex, nonadditive phenomena like through-space polarization. (32,33) The accuracy of these models is critical, as even small inaccuracies can result in significant errors in the predicted properties, like binding affinities or reaction rates. (34) The importance of force fields has only increased as computational methods have become more sophisticated. For example, the advent of hybrid quantum mechanics/molecular mechanics (QM/MM) approaches, which combine QM models with classical force fields, (35) has rendered possible the theoretical investigation of chemical reactions in large systems, such as catalysis in enzymes, where the reactive site is treated quantum mechanically, while the remainder of the system is described using a classical force field. In turn, this methodology can be harnessed to improve at a minimal additional cost the description of certain interactions, otherwise poorly modeled by standard pairwise additive force fields, like cation−π interactions. (36) To a certain extent, the success of QM/MM methods depends critically on the accuracy of the force field utilized to describe the nonreactive region of the system. Furthermore, the development of force fields tailored for specific types of systems, like membrane environments (37,38) or carbohydrates, (39) has allowed more accurate, more focused, and, in a sense, more realistic simulations to be performed. These specialized force fields take into account the unique interactions and structural features of different molecular systems and provide a more precise representation of their behavior in a true biological context. ML, again, has further enhanced the development and application of force fields by allowing more reliable and flexible models to be designed. (40) ML strategies can be employed to refine force-field parameters, ensuring that they better represent experimental data or high-level QM calculations. Additionally, ML-driven force fields, such as artificial-neural-network (ANN) potentials, have emerged as powerful tools that can capture complex interactions with a level of detail that traditional force fields struggle to achieve. (41) These advancements are particularly important in the study of systems where traditional force fields may fall short, like in the modeling of complex chemical environments.The ever-evolving landscape of theoretical and computational chemistry, along with biophysics, owes much of its progress to both the continuous advancements in software tools and the development of accurate force fields. These twin pillars have fundamentally reshaped the way researchers approach complex molecular systems, enabling more precise simulations and predictions. The integration of sophisticated algorithms with high-performance computing has allowed scientists to tackle previously intractable problems, from understanding intricate biomolecular interactions to designing novel materials with specific properties. This Virtual Special Issue seeks to highlight in a collection of articles the latest innovations in both software and force field development, underscoring their critical role in pushing the boundaries of computational science.On the software front, Eastman et al. report the latest version of OpenMM, which integrates ML potentials, enabling PyTorch models to compute forces and energy in simulations, improving accuracy at minimal cost increase (DOI: 10.1021/acs.jpcb.3c06662). Gissinger et al. propose a new framework in LAMMPS, which enhances compatibility with supporting software, simplifies bonded force field representation, and improves usability with a variety of tools, streamlining molecular-simulation workflows (DOI: 10.1021/acs.jpcb.3c08419). Kaiser et al. present a new method in the RAPTOR software for LAMMPS to enable efficient long-time-scale reactive simulations, with GPU acceleration and new collective variables to model phenomena like proton transport and water wire formation (DOI: 10.1021/acs.jpcb.4c01987). Jung et al. report the latest version of GENESIS for multiscale MD simulations, with enhanced sampling algorithms, and integrated free-energy calculations (DOI: 10.1021/acs.jpcb.4c02096). Schäfer and Keller report an implementation of Girsanov reweighting in OpenMM, allowing real-time path reweighting in MD simulations, improving enhanced sampling techniques, and enabling unbiased dynamics recovery without altering the MD model (DOI: 10.1021/acs.jpcb.4c01702). Giese et al. integrate AMBER with xtb quantum models and DeePMD-kit ML potentials to create accurate QM/MM−ΔMLP force fields for diverse MD and free-energy calculations (DOI: 10.1021/acs.jpcb.4c01466). Hwang et al. describe the evolution of CHARMM with new simulation engines, user-friendly interfaces, and comprehensive methods across quantum, atomistic, and coarse-grained levels, supporting diverse biomolecular research (DOI: 10.1021/acs.jpcb.4c04100). Fiorin et al. report the latest advanced features available in the Colvars library from ML optimization, interactive descriptor exploration, to various biasing algorithms, now integrated into major MD packages and ensured by continuous testing (DOI: 10.1021/acs.jpcb.4c05604). Wesołowski et al. propose a three-layered multicenter ONIOM approach to characterize the BPTI folding pathway, highlighting the importance of disulfide bonds and balancing accuracy with computational efficiency (DOI: 10.1021/acs.jpcb.4c00104). Ikizawa et al. present an enhanced sampling method using parallel MD simulations and a toolkit to enable its customizable execution and analysis across various platforms (DOI: 10.1021/acs.jpcb.4c01271). Shi et al. show how transfer learning in ML models for classical MD enhances accuracy and efficiency by integrating multifidelity data, significantly reducing computational costs while maintaining predictive performance (DOI: 10.1021/acs.jpca.4c00750). Matúška et al. report how the optimization of the cutoff parameter in SchNetPack improves top docking score predictions more effectively than data sampling adjustments (DOI: 10.1021/acs.jpcb.4c00296). Last, Wang et al. have developed a workflow coined "PepBinding" to improve protein-peptide binding predictions by combining peptide docking with enhanced-sampling simulations, which increases significantly the accuracy of the model within 200 ns of simulation (DOI: 10.1021/acs.jpcb.4c02047).On the force-field front, Jorgensen et al. report an update of the OPLS/2020 force field, which improves accuracy in modeling unsaturated hydrocarbons, alcohols, and ethers, with reduced errors in liquid properties and hydration free energies (DOI: 10.1021/acs.jpcb.3c06602). Salom-Català et al. have developed a force field that facilitates large-scale simulations of propane dehydrogenation on Pt surfaces, accurately modeling reaction mechanisms and catalyst reactivity across different surface topologies and temperatures (DOI: 10.1021/acs.jpcc.3c07126). Wang et al. present the AMOEBA+ANN hybrid model, which combines AMOEBA for long-range interactions with an ANN potential for local covalent contributions, offering improved accuracy in molecular simulations as molecule size increases (DOI: 10.1021/acs.jpcb.3c08166). Lucker et al. report an update of the C36UAr united-atom force field, which includes sphingolipids and offers improved accuracy and reduced simulation time compared to all-atom models, while maintaining consistency with phospholipid simulations (DOI: 10.1021/acs.jpcb.4c01404). Kumar and MacKerell have developed FFParam-v2 to improve CHARMM force-field optimization with new capabilities for condensed-phase data, Lennard-Jones fitting, and automated workflows (DOI: 10.1021/acs.jpcb.4c01314). Stroet et al. evaluate different GROMOS protein force field parameters using a curated test set, revealing small, often compensatory differences in various structural metrics, which underscores the need for comprehensive validation frameworks (DOI: 10.1021/acs.jpcb.3c08469). Wang et al. put forth a hybrid physical/graph NN (GNN) model offering a faster and accurate alternative to AM1-BCC for assigning atomic partial charges, scalable for large systems and integrated into the EspalomaCharge package (DOI: 10.1021/acs.jpca.4c01287). Yuan et al. introduce a GNN-based force field to enhance the accuracy of MD simulations for chemical reactions, enabling precise study of reactions like Claisen rearrangement and carbonyl insertion at lower computational costs (DOI: 10.1021/acs.jpca.4c01267). Wang et al. review advances in the Open Force Field initiative in developing next-generation biomolecular force fields, emphasizing its open approach to software, data, and collaborative research (DOI: 10.1021/acs.jpcb.4c01558). Jin et al. introduce a training pipeline for machine-learned force fields that enables stable simulations over tens of nanoseconds, surpassing traditional methods (10.1021/acs.jpca.4c01546). Raguette et al. improve RNA simulations by adjusting CH···O interaction parameters in force fields, enhancing stability and accuracy across various RNA systems (DOI: 10.1021/acs.jpcb.4c01910). Finally, Cantero et al. show that the coarse-grained SIRAH force field can be effectively used with the NAMD code, including its GPU-accelerated version, enhancing simulation throughput across different systems (DOI: 10.1021/acs.jpcb.4c03278).These contributions showcase how cutting-edge software is being leveraged to address new challenges, while also emphasizing the crucial role of force fields in ensuring the accuracy and reliability of the simulations performed with these tools. As the field continues to evolve, particularly with the advent of ML techniques, these updates are more important than ever in maintaining the relevance and utility of computational methods. We hope this collection of research articles will serve as both a valuable resource and an inspiration for future developments in the field.Author InformationClick to copy section linkSection link copied!Corresponding AuthorChristophe Chipot - Laboratoire International Associé Centre National de la Recherche Scientifique et University of Illinois at Urbana−Champaign, UMR n°7019, Université de Lorraine, 54506 Vandoeuvre-lès-Nancy cedex, France; Department of Physics and Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana−Champaign, Urbana, Illinois 61801, United States; Department of Biochemistry and Molecular Biology, The University of Chicago, Chicago, Illinois 60637, United States; https://orcid.org/0000-0002-9122-1698; Email: [email protected]NotesViews expressed in this preface are those of the author and not necessarily the views of the ACS.ReferencesClick to copy section linkSection link copied! This article references 41 other publications. 1Boys, S. F.; Cook, G. B.; Reeves, C. M.; Shavitt, I. Automatic Fundamental Calculations of Molecular Structure. Nature 1956, 178, 1207– 1209, DOI: 10.1038/1781207a0 Google ScholarThere is no corresponding record for this reference.2Alder, B. J.; Wainwright, T. E. Phase Transition for a Hard Sphere System. J. Chem. Phys. 1957, 27, 1208– 1209, DOI: 10.1063/1.1743957 Google Scholar2Phase transition for a hard-sphere systemAlder, B. J.; Wainwright, T. E.Journal of Chemical Physics (1957), 27 (), 1208CODEN: JCPSA6; ISSN:0021-9606. The method consisting of solving exactly the simultaneous classical equations of motion of several hundred particles by means of fast electronic computers was used. The results of the calcn. are given in a graph in the paper covered by the preceding abstr. Two sep. and overlapping branches of the curve were obtained. >> More from SciFinder ®https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaG1cXitlymsg%253D%253D&md5=f9236d66761cb65c4f73b4dc7a2a36733Weiner, P. K.; Kollman, P. A. Amber: Assisted Model Building with Energy Refinement. A General Program for Modeling Molecules and Their Interactions. J. Comput. Chem. 1981, 2, 287– 303, DOI: 10.1002/jcc.540020311 Google Scholar3AMBER: assisted model building with energy refinement. A general program for modeling molecules and their interactionsWeiner, Paul K.; Kollman, Peter A.Journal of Computational Chemistry (1981), 2 (3), 287-303CODEN: JCCHDD; ISSN:0192-8651. A computer program was developed to build models of mols. and calc. their interactions using empirical energy approaches. The program is sufficiently flexible and general to allow modeling of small mols., as well as polymers. The conformation of actinomycin D was studied for illustration. The rotational isomerism about the D-Val-, L-Pro, and L-Pro-Sar amide bonds was studied. The energy and structure of the Sobell model and the x-ray structure of actinomycin D were compared. >> More from SciFinder ®https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaL3MXltFCltLs%253D&md5=0921bad2996e8a57a17c4129053405b94Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J.; Swaminathan, S.; Karplus, M. Charmm: A Program for Macromolecular Energy, Minimization, and Dynamics Calculations. J. Comput. Chem. 1983, 4, 187– 217, DOI: 10.1002/jcc.540040211 Google Scholar4CHARMM: a program for macromolecular energy, minimization, and dynamics calculationsBrooks, Bernard R.; Bruccoleri, Robert E.; Olafson, Barry D.; States, David J.; Swaminathan, S.; Karplus, MartinJournal of Computational Chemistry (1983), 4 (2), 187-217CODEN: JCCHDD; ISSN:0192-8651. CHARMM (Chem. at HARvard Macromol. Mechanics) is a highly flexible computer program which uses empirical energy functions to model macromol. systems. The program can read or model build structures, energy minimize them by first- or second-deriv. techniques, perform a normal mode or mol. dynamics simulation, and analyze the structural, equil., and dynamic properties detd. in these calcns. The operations that CHARMM can perform are described, and some implementation details are given. A set of parameters for the empirical energy function and a sample run are included. >> More from SciFinder ®https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaL3sXit1aiu7w%253D&md5=bd639b4299ac9934f4497c1a9fe750d25Van Gunsteren, W. F.; Berendsen, H. J. C. Groningen Molecular Simulation (Gromos); University of Groningen: Groningen, 1987.Google ScholarThere is no corresponding record for this reference.6Duan, Y.; Kollman, P. A. Pathways to a Protein Folding Intermediate Observed in a 1-Microsecond Simulation in Aqueous Solution. Science 1998, 282, 740– 744, DOI: 10.1126/science.282.5389.740 Google Scholar6Pathways to a protein folding intermediate observed in a 1-microsecond simulation in aqueous solutionDuan, Yong; Kollman, Peter A.Science (Washington, D. C.) (1998), 282 (5389), 740-744CODEN: SCIEAS; ISSN:0036-8075. (American Association for the Advancement of Science) An implementation of classical mol. dynamics on parallel computers of increased efficiency has enabled a simulation of protein folding with explicit representation of water for 1 μs, about two orders of magnitude longer than the longest simulation of a protein in water reported to date. Starting with an unfolded state of villin headpiece subdomain, hydrophobic collapse and helix formation occur in an initial phase, followed by conformational readjustments. A marginally stable state, which has a lifetime of about 150 ns, a favorable solvation free energy, and shows significant resemblance to the native structure, is obsd.; two pathways to this state have been found. >> More from SciFinder ®https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK1cXmvFOqtb0%253D&md5=e3dde8454d4dbcc47f2a69f4c00e4ded7Gabdoulline, R. R.; Wade, R. C. Simulation of the Diffusional Association of Barnase and Barstar. Biophys. J. 1997, 72, 1917– 1929, DOI: 10.1016/S0006-3495(97)78838-6 Google Scholar7Simulation of the diffusional association of Barnase and BarstarGabdoulline, R. R.; Wade, R. C.Biophysical Journal (1997), 72 (5), 1917-1929CODEN: BIOJAU; ISSN:0006-3495. (Biophysical Society) The rate of protein assocn. places an upper limit on the response time due to protein interactions, which, under certain circumstances, can be diffusion-controlled. Simulations of model proteins show that diffusion-limited assocn. rates are ∼106-107 M-1 s-1 in the absence of long-range forces. The measured assocn. rates of barnase and barstar are 108-109 M-1 s-1 at 50 mM ionic strength, and depend on ionic strength, implying that their assocn. is electrostatically facilitated. We report Brownian dynamics simulations of the diffusional assocn. of barnase and barstar to compute assocn. rates and their dependence on ionic strength and protein mutation. Crucial to the ability to reproduce exptl. rates is the definition of encounter complex formation at the endpoint of diffusional motion. Simple definitions, such as a required root mean square (RMS) distance to the fully bound position, fail to explain the large influence of some mutations on assocn. rates. Good agreement with expts. could be obtained if satisfaction of two intermol. residue contacts was required for encounter complex formation. In the encounter complexes, barstar tends to be shifted from its position in the bound complex toward the guanine-binding loop on barnase. >> More from SciFinder ®https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADyaK2sXivVKqtbc%253D&md5=17588acd2fbf969d0ad7c6a1e388d24b8Garcia-Viloca, M.; Gao, J.; Karplus, M.; Truhlar, D. G. How Enzymes Work: Analysis by Modern Rate Theory and Computer Simulations. Science 2004, 303, 186– 195, DOI: 10.1126/science.1088172 Google Scholar8How enzymes work: Analysis by modern rate theory and computer simulationsGarcia-Viloca, Mireia; Gao, Jiali; Karplus, Martin; Truhlar, Donald G.Science (Washington, DC, United States) (2004), 303 (5655), 186-195CODEN: SCIEAS; ISSN:0036-8075. (American Association for the Advancement of Science) A review and discussion. Advances in transition state theory and computer simulations are providing new insights into the sources of enzyme catalysis. Both the lowering of the activation free energy and changes in the generalized transmission coeff. (recrossing of the transition state, tunneling, and nonequil. contributions) can play a role. A framework for understanding these effects is presented, and the contributions of the different factors, as illustrated by specific enzymes, are identified and quantified by computer simulations. The resulting understanding of enzyme catalysis is used to comment on alternative proposals of how enzymes work. >> More from SciFinder ®https://chemport.cas.org/services/resolver?origin=ACS&resolution=options&coi=1%3ACAS%3A528%3ADC%252BD2cXhtlGhug%253D%253D&md5=35766752530af526c8bdc99774a8e9da9Abraham, M. J.; Murtola, T.; Schulz, R.; Páll, S.; Smith, J. C.; Hess, B.; Lindahl, E. Gromacs: High Performance Molecular Simulations through Multi-Level Parallelism from Laptops to Supercomputers. SoftwareX 2015, 1–2, 19– 25, DOI: 10.1016/j.softx.2015.06.001 Google ScholarThere is no corresponding record for this reference.10Pande, V.; Eastman, P. Openmm: A Hardware-Independent Framework for Molecular Simulations ; 2010; pp 34– 39.Google ScholarThere is no corresponding record for this reference.11Jung, J.; Mori, T.; Kobayashi, C.; Matsunaga, Y.; Yoda, T.; Feig, M.; Sugita, Y. Genesis: A Hybrid-Parallel and Multi-Scale Molecular Dynamics Simulator with Enhanced Sampling Algorithms for Biomolecular and Cellular Simulations. WIREs Computational Molecular Science 2015, 5, 310– 323, DOI: 10.1002/wcms.1220 Google ScholarThere is no corresponding record for this reference.12Phillips, J. C.; Hardy, D. J.; Maia, J. D. C.; Stone, J. E.; Ribeiro, J. V.; Bernardi, R. C.; Buch, R.; Fiorin, G.; Hénin, J.; Jiang, W. Scalable Molecular Dynamics on Cpu and Gpu Architectures with Namd. J. Chem. Phys. 2020, 153, 044130, DOI: 10.1063/5.0014475 Google Scholar12Scalable molecular dynamics on CPU and GPU architectures with NAMDPhillips, James C.; Hardy, David J.; Maia, Julio D. C.; Stone, John E.; Ribeiro, Joao V.; Bernardi, Rafael C.; Buch, Ronak; Fiorin, Giacomo; Henin, Jerome; Jiang, Wei; McGreevy, Ryan; Melo, Marcelo C. R.; Radak, Brian K.; Skeel, Robert D.; Singharoy, Abhishek; Wang, Yi; Roux, Benoit; Aksimentiev, Aleksei; Luthey-Schulten, Zaida; Kale, Laxmikant V.; Schulten, Klaus; Chipot, Christophe; Tajkhorshid, EmadJournal of Chemical Physics (2020), 153 (4), 044130CODEN: JCPSA6; ISSN:0021-9606. (American Institute of Physics) A review. NAMD is a mol. dynamics program designed for high-performance simulations of very large biol. objects on CPU- and GPU-based architectures. NAMD offers scalable performance on petascale parallel supercomputers consisting of hundreds of thousands of cores, as well as on inexpensive commodity clusters commonly found in academic environments. It is written in C++ and leans on Charm++ parallel objects for optimal performance on low-latency architectures. NAMD is a versatile, multipurpose code that gathers state-of-the-art algorithms to carry out simulations in apt thermodn. ensembles, using the widely popular CHARMM, AMBER, OPLS, and GROMOS biomol. force fields. Here, the authors review the main features of NAMD that allow both equil. and enhanced-sampling mol. dynamics simulations with numerical efficiency. The authors describe the underlying concepts used by NAMD and their implementation, most notably for handling long-range electrostatics; controlling the temp., pressure, and pH; applying external potentials on tailored grids; leveraging massively parallel resources in multiple-copy simulations; and hybrid quantum-mech./mol.-mech. descriptions. The authors detail the variety of options offered by NAMD for enhanced-sampling simulations aimed at detg. free-energy differences of either alchem. or geometrical transformations and outline their applicability to specific problems. Last, the roadmap for the development of NAMD and the authors' current efforts toward achieving optimal performance on GPU-based architectures, for pushing back the limitat
Referência(s)