The ugly, bad, and good stories of large-scale biomolecular simulations

Advancing biomolecular simulation through exascale HPC, AI and quantum computing

Biomolecular simulation can act as both a digital microscope and a crystal ball; offering the potential for a deeper understanding of experimental observations whilst also presenting a forward-looking avenue for the in silico design and evaluation of hitherto unsynthesized compounds. Indeed, as the intricacy of our scientific inquiries has grown, so too has the computational prowess we seek to deploy in our pursuit of answers. As we enter the Exascale era, this mini-review surveys the computational landscape from both the point of view of the development of new and ever more powerful systems, and the simulations that are run on them. Moreover, as we stand on the cusp of a transformative phase in computational biology, this article offers a contemplative glance into the future, speculating on the profound implications of artificial intelligence and quantum computing for large-scale biomolecular simulations.

Embracing exascale computing in nucleic acid simulations

This mini-review reports the recent advances in biomolecular simulations, particularly for nucleic acids, and provides the potential effects of the emerging exascale computing on nucleic acid simulations, emphasizing the need for advanced computational strategies to fully exploit this technological frontier. Specifically, we introduce recent breakthroughs in computer architectures for large-scale biomolecular simulations and review the simulation protocols for nucleic acids regarding force fields, enhanced sampling methods, coarse-grained models, and interactions with ligands. We also explore the integration of machine learning methods into simulations, which promises to significantly enhance the predictive modeling of biomolecules and the analysis of complex data generated by the exascale simulations. Finally, we discuss the challenges and perspectives for biomolecular simulations as we enter the dawning exascale computing era.

Modeling membranes in situ

Molecular dynamics simulations of cellular membranes have come a long way-from simple model lipid bilayers to multicomponent systems capturing the crowded and complex nature of real cell membranes. In this opinionated minireview, we discuss the current challenge to simulate the dynamics of membranes in their native environment, in situ, with the prospect of reaching the level of whole cells and cell organelles using an integrative modeling framework.

Pathways to a Shiny Future: Building the Foundation for Computational Physical Chemistry and Biophysics in 2050

In the last quarter-century, the field of molecular dynamics (MD) has undergone a remarkable transformation, propelled by substantial enhancements in software, hardware, and underlying methodologies. In this Perspective, we contemplate the future trajectory of MD simulations and their possible look at the year 2050. We spotlight the pivotal role of artificial intelligence (AI) in shaping the future of MD and the broader field of computational physical chemistry. We outline critical strategies and initiatives that are essential for the seamless integration of such technologies. Our discussion delves into topics like multiscale modeling, adept management of ever-increasing data deluge, the establishment of centralized simulation databases, and the autonomous refinement, cross-validation, and self-expansion of these repositories. The successful implementation of these advancements requires scientific transparency, a cautiously optimistic approach to interpreting AI-driven simulations and their analysis, and a mindset that prioritizes knowledge-motivated research alongside AI-enhanced big data exploration. While history reminds us that the trajectory of technological progress can be unpredictable, this Perspective offers guidance on preparedness and proactive measures, aiming to steer future advancements in the most beneficial and successful direction.

There and back again: bridging meso- and nano-scales to understand lipid vesicle patterning

We describe a complete methodology to bridge the scales between nanoscale molecular dynamics and (micrometer) mesoscale Monte Carlo simulations in lipid membranes and vesicles undergoing phase separation, in which curving molecular species are furthermore embedded. To go from the molecular to the mesoscale, we notably appeal to physical renormalization arguments enabling us to rigorously infer the mesoscale interaction parameters from its molecular counterpart. We also explain how to deal with the physical timescales at stake at the mesoscale. Simulating the as-obtained mesoscale system enables us to equilibrate the long wavelengths of the vesicles of interest, up to the vesicle size. Conversely, we then backmap from the meso- to the nano-scale, which enables us to equilibrate in turn the short wavelengths down to the molecular length-scales. By applying our approach to the specific situation of patterning a vesicle membrane, we show that macroscopic membranes can thus be equilibrated at all length-scales in achievable computational time offering an original strategy to address the fundamental challenge of timescale in simulations of large bio-membrane systems.

A Gauss-Newton method for iterative optimization of memory kernels for generalized Langevin thermostats in coarse-grained molecular dynamics simulations

In molecular dynamics simulations, dynamically consistent coarse-grained (CG) models commonly use stochastic thermostats to model friction and fluctuations that are lost in a CG description. While Markovian, i.e., time-local, formulations of such thermostats allow for an accurate representation of diffusivities/long-time dynamics, a correct description of the dynamics on all time scales generally requires non-Markovian, i.e., non-time-local, thermostats. These thermostats typically take the form of a Generalized Langevin Equation (GLE) determined by a memory kernel. In this work, we use a Markovian embedded formulation of a position-independent GLE thermostat acting independently on each CG degree of freedom. Extracting the memory kernel of this CG model from atomistic reference data requires several approximations. Therefore, this task is best understood as an inverse problem. While our recently proposed approximate Newton scheme allows for the iterative optimization of memory kernels (IOMK), Markovian embedding remained potentially error-prone and computationally expensive. In this work, we present an IOMK-Gauss-Newton scheme (IOMK-GN) based on IOMK that allows for the direct parameterization of a Markovian embedded model.

Evaluating Polarizable Biomembrane Simulations against Experiments

Owing to the increase of available computational capabilities and the potential for providing a more accurate description, polarizable molecular dynamics force fields are gaining popularity in modeling biomolecular systems. It is, however, crucial to evaluate how much precision is truly gained with increasing cost and complexity of the simulation. Here, we leverage the NMRlipids open collaboration and Databank to assess the performance of available polarizable lipid models─the CHARMM-Drude and the AMOEBA-based parameters─against high-fidelity experimental data and compare them to the top-performing nonpolarizable models. While some improvement in the description of ion binding to membranes is observed in the most recent CHARMM-Drude parameters, and the conformational dynamics of AMOEBA-based parameters are excellent, the best nonpolarizable models tend to outperform their polarizable counterparts for each property we explored. The identified shortcomings range from inaccuracies in describing the conformational space of lipids to excessively slow conformational dynamics. Our results provide valuable insights for the further refinement of polarizable lipid force fields and for selecting the best simulation parameters for specific applications.

MDverse: Shedding Light on the Dark Matter of Molecular Dynamics Simulations

The rise of open science and the absence of a global dedicated data repository for molecular dynamics (MD) simulations has led to the accumulation of MD files in generalist data repositories, constituting the dark matter of MD - data that is technically accessible, but neither indexed, curated, or easily searchable. Leveraging an original search strategy, we found and indexed about 250,000 files and 2,000 datasets from Zenodo, Figshare and Open Science Framework. With a focus on files produced by the Gromacs MD software, we illustrate the potential offered by the mining of publicly available MD data. We identified systems with specific molecular composition and were able to characterize essential parameters of MD simulation such as temperature and simulation length, and could identify model resolution, such as all-atom and coarse-grain. Based on this analysis, we inferred metadata to propose a search engine prototype to explore the MD data. To continue in this direction, we call on the community to pursue the effort of sharing MD data, and to report and standardize metadata to reuse this valuable matter.

Construction of Multiscale Dissipative Particle Dynamics (DPD) Models from Other Coarse-Grained Models

We present a general scheme for converting coarse-grained models into Dissipative Particle Dynamics (DPD) models. We build the corresponding DPD models by analogy with the de novo DPD coarse-graining scheme suggested by Groot and Warren (J. Chem. Phys., 1997). Electrostatic interactions between charged DPD particles are represented though the addition of a long-range Slater Coulomb potential as suggested by González-Melchor et al. (J. Chem. Phys., 2006). The construction is illustrated by converting MARTINI models for various proteins into a DPD representation, but it not restricted to the usual potential form in the MARTINI model-viz., Lennard-Jones potentials. We further extended the DPD scheme away from the typical use of homogeneous particle sizes, therefore faithfully representing the variations in the particle sizes seen in the underlying MARTINI model. The accuracy of the resulting construction of our generalized DPD models with respect to several structural observables has been benchmarked favorably against all-atom and MARTINI models for a selected set of peptides and proteins, and variations in the scales of the coarse-graining of the water solvent.

ThermoPCD: a database of molecular dynamics trajectories of antibody-antigen complexes at physiologic and fever-range temperatures

Progression of various cancers and autoimmune diseases is associated with changes in systemic or local tissue temperatures, which may impact current therapies. The role of fever and acute inflammation-range temperatures on the stability and activity of antibodies relevant for cancers and autoimmunity is unknown. To produce molecular dynamics (MD) trajectories of immune complexes at relevant temperatures, we used the Research Collaboratory for Structural Bioinformatics (RCSB) database to identify 50 antibody:antigen complexes of interest, in addition to single antibodies and antigens, and deployed Groningen Machine for Chemical Simulations (GROMACS) to prepare and run the structures at different temperatures for 100-500 ns, in single or multiple random seeds. MD trajectories are freely available. Processed data include Protein Data Bank outputs for all files obtained every 50 ns, and free binding energy calculations for some of the immune complexes. Protocols for using the data are also available. Individual datasets contain unique DOIs. We created a web interface, ThermoPCD, as a platform to explore the data. The outputs of ThermoPCD allow the users to relate thermally-dependent changes in epitopes:paratopes interfaces to their free binding energies, or against own experimentally derived binding affinities. ThermoPCD is a free to use database of immune complexes' trajectories at different temperatures that does not require registration and allows for all the data to be available for download. Database URL: https://sites.google.com/view/thermopcd/home.

All-Atom Biomolecular Simulation in the Exascale Era

Exascale supercomputers have opened the door to dynamic simulations, facilitated by AI/ML techniques, that model biomolecular motions over unprecedented length and time scales. This new capability holds the potential to revolutionize our understanding of fundamental biological processes. Here we report on some of the major advances that were discussed at a recent CECAM workshop in Pisa, Italy, on the topic with a primary focus on atomic-level simulations. First, we highlight examples of current large-scale biomolecular simulations and the future possibilities enabled by crossing the exascale threshold. Next, we discuss challenges to be overcome in optimizing the usage of these powerful resources. Finally, we close by listing several grand challenge problems that could be investigated with this new computer architecture.

Overlay databank unlocks data-driven analyses of biomolecules for all

Tools based on artificial intelligence (AI) are currently revolutionising many fields, yet their applications are often limited by the lack of suitable training data in programmatically accessible format. Here we propose an effective solution to make data scattered in various locations and formats accessible for data-driven and machine learning applications using the overlay databank format. To demonstrate the practical relevance of such approach, we present the NMRlipids Databank-a community-driven, open-for-all database featuring programmatic access to quality-evaluated atom-resolution molecular dynamics simulations of cellular membranes. Cellular membrane lipid composition is implicated in diseases and controls major biological functions, but membranes are difficult to study experimentally due to their intrinsic disorder and complex phase behaviour. While MD simulations have been useful in understanding membrane systems, they require significant computational resources and often suffer from inaccuracies in model parameters. Here, we demonstrate how programmable interface for flexible implementation of data-driven and machine learning applications, and rapid access to simulation data through a graphical user interface, unlock possibilities beyond current MD simulation and experimental studies to understand cellular membranes. The proposed overlay databank concept can be further applied to other biomolecules, as well as in other fields where similar barriers hinder the AI revolution.

Development of nano-delivery systems for loaded bioactive compounds: using molecular dynamics simulations

Over the past decade, a remarkable surge in the development of functional nano-delivery systems loaded with bioactive compounds for healthcare has been witnessed. Notably, the demanding requirements of high solubility, prolonged circulation, high tissue penetration capability, and strong targeting ability of nanocarriers have posed interdisciplinary research challenges to the community. While extensive experimental studies have been conducted to understand the construction of nano-delivery systems and their metabolic behavior in vivo, less is known about these molecular mechanisms and kinetic pathways during their metabolic process in vivo, and lacking effective means for high-throughput screening. Molecular dynamics (MD) simulation techniques provide a reliable tool for investigating the design of nano-delivery carriers encapsulating these functional ingredients, elucidating the synthesis, translocation, and delivery of nanocarriers. This review introduces the basic MD principles, discusses how to apply MD simulation to design nanocarriers, evaluates the ability of nanocarriers to adhere to or cross gastrointestinal mucosa, and regulates plasma proteins in vivo. Moreover, we presented the critical role of MD simulation in developing delivery systems for precise nutrition and prospects for the future. This review aims to provide insights into the implications of MD simulation techniques for designing and optimizing nano-delivery systems in the healthcare food industry.

Adaptive Ensemble Refinement of Protein Structures in High Resolution Electron Microscopy Density Maps with Radical Augmented Molecular Dynamics Flexible Fitting

Recent advances in cryo-electron microscopy (cryo-EM) have enabled modeling macromolecular complexes that are essential components of the cellular machinery. The density maps derived from cryo-EM experiments are often integrated with manual, knowledge-driven or artificial intelligence-driven and physics-guided computational methods to build, fit, and refine molecular structures. Going beyond a single stationary-structure determination scheme, it is becoming more common to interpret the experimental data with an ensemble of models that contributes to an average observation. Hence, there is a need to decide on the quality of an ensemble of protein structures on-the-fly while refining them against the density maps. We introduce such an adaptive decision-making scheme during the molecular dynamics flexible fitting (MDFF) of biomolecules. Using RADICAL-Cybertools, the new RADICAL augmented MDFF implementation (R-MDFF) is examined in high-performance computing environments for refinement of two prototypical protein systems, adenylate kinase and carbon monoxide dehydrogenase. For these test cases, use of multiple replicas in flexible fitting with adaptive decision making in R-MDFF improves the overall correlation to the density by 40% relative to the refinements of the brute-force MDFF. The improvements are particularly significant at high, 2-3 Å map resolutions. More importantly, the ensemble model captures key features of biologically relevant molecular dynamics that are inaccessible to a single-model interpretation. Finally, the pipeline is applicable to systems of growing sizes, which is demonstrated using ensemble refinement of capsid proteins from the chimpanzee adenovirus. The overhead for decision making remains low and robust to computing environments. The software is publicly available on GitHub and includes a short user guide to install R-MDFF on different computing environments, from local Linux-based workstations to high-performance computing environments.

Fostering discoveries in the era of exascale computing: How the next generation of supercomputers empowers computational and experimental biophysics alike

Over a century ago, physicists started broadly relying on theoretical models to guide new experiments. Soon thereafter, chemists began doing the same. Now, biological research enters a new era when experiment and theory walk hand in hand. Novel software and specialized hardware became essential to understand experimental data and propose new models. In fact, current petascale computing resources already allow researchers to reach unprecedented levels of simulation throughput to connect in silico and in vitro experiments. The reduction in cost and improved access allowed a large number of research groups to adopt supercomputing resources and techniques. Here, we outline how large-scale computing has evolved to expand decades-old research, spark new research efforts, and continuously connect simulation and observation. For instance, multiple publicly and privately funded groups have dedicated extensive resources to develop artificial intelligence tools for computational biophysics, from accelerating quantum chemistry calculations to proposing protein structure models. Moreover, advances in computer hardware have accelerated data processing from single-molecule experimental observations and simulations of chemical reactions occurring throughout entire cells. The combination of software and hardware has opened the way for exascale computing and the production of the first public exascale supercomputer, Frontier, inaugurated by the Oak Ridge National Laboratory in 2022. Ultimately, the popularization and development of computational techniques and the training of researchers to use them will only accelerate the diversification of tools and learning resources for future generations.

Building Asymmetric Lipid Bilayers for Molecular Dynamics Simulations: What Methods Exist and How to Choose One?

The compositional asymmetry of biological membranes has attracted significant attention over the last decade. Harboring more differences from symmetric membranes than previously appreciated, asymmetric bilayers have proven quite challenging to study with familiar concepts and techniques, leaving many unanswered questions about the reach of the asymmetry effects. One particular area of active research is the computational investigation of composition- and number-asymmetric lipid bilayers with molecular dynamics (MD) simulations. Offering a high level of detail into the organization and properties of the simulated systems, MD has emerged as an indispensable tool in the study of membrane asymmetry. However, the realization that results depend heavily on the protocol used for constructing the asymmetric bilayer models has sparked an ongoing debate about how to choose the most appropriate approach. Here we discuss the underlying source of the discrepant results and review the existing methods for creating asymmetric bilayers for MD simulations. Considering the available data, we argue that each method is well suited for specific applications and hence there is no single best approach. Instead, the choice of a construction protocol-and consequently, its perceived accuracy-must be based primarily on the scientific question that the simulations are designed to address.

Analyzing the Geometry and Dynamics of Viral Structures: A Review of Computational Approaches Based on Alpha Shape Theory, Normal Mode Analysis, and Poisson-Boltzmann Theories

The current SARS-CoV-2 pandemic highlights our fragility when we are exposed to emergent viruses either directly or through zoonotic diseases. Fortunately, our knowledge of the biology of those viruses is improving. In particular, we have more and more structural information on virions, i.e., the infective form of a virus that includes its genomic material and surrounding protective capsid, and on their gene products. It is important to have methods that enable the analyses of structural information on such large macromolecular systems. We review some of those methods in this paper. We focus on understanding the geometry of virions and viral structural proteins, their dynamics, and their energetics, with the ambition that this understanding can help design antiviral agents. We discuss those methods in light of the specificities of those structures, mainly that they are huge. We focus on three of our own methods based on the alpha shape theory for computing geometry, normal mode analyses to study dynamics, and modified Poisson-Boltzmann theories to study the organization of ions and co-solvent and solvent molecules around biomacromolecules. The corresponding software has computing times that are compatible with the use of regular desktop computers. We show examples of their applications on some outer shells and structural proteins of the West Nile Virus.

Machines on Genes through the Computational Microscope

Macromolecular machines acting on genes are at the core of life's fundamental processes, including DNA replication and repair, gene transcription and regulation, chromatin packaging, RNA splicing, and genome editing. Here, we report the increasing role of computational biophysics in characterizing the mechanisms of "machines on genes", focusing on innovative applications of computational methods and their integration with structural and biophysical experiments. We showcase how state-of-the-art computational methods, including classical and ab initio molecular dynamics to enhanced sampling techniques, and coarse-grained approaches are used for understanding and exploring gene machines for real-world applications. As this review unfolds, advanced computational methods describe the biophysical function that is unseen through experimental techniques, accomplishing the power of the "computational microscope", an expression coined by Klaus Schulten to highlight the extraordinary capability of computer simulations. Pushing the frontiers of computational biophysics toward a pragmatic representation of large multimegadalton biomolecular complexes is instrumental in bridging the gap between experimentally obtained macroscopic observables and the molecular principles playing at the microscopic level. This understanding will help harness molecular machines for medical, pharmaceutical, and biotechnological purposes.

Molecular architecture and dynamics of SARS-CoV-2 envelope by integrative modeling

Despite tremendous efforts, the exact structure of SARS-CoV-2 and related betacoronaviruses remains elusive. SARS-CoV-2 envelope is a key structural component of the virion that encapsulates viral RNA. It is composed of three structural proteins, spike, membrane (M), and envelope, which interact with each other and with the lipids acquired from the host membranes. Here, we developed and applied an integrative multi-scale computational approach to model the envelope structure of SARS-CoV-2 with near atomistic detail, focusing on studying the dynamic nature and molecular interactions of its most abundant, but largely understudied, M protein. The molecular dynamics simulations allowed us to test the envelope stability under different configurations and revealed that the M dimers agglomerated into large, filament-like, macromolecular assemblies with distinct molecular patterns. These results are in good agreement with current experimental data, demonstrating a generic and versatile approach to model the structure of a virus de novo.

Molecular dynamics simulation of an entire cell

The ultimate microscope, directed at a cell, would reveal the dynamics of all the cell's components with atomic resolution. In contrast to their real-world counterparts, computational microscopes are currently on the brink of meeting this challenge. In this perspective, we show how an integrative approach can be employed to model an entire cell, the minimal cell, JCVI-syn3A, at full complexity. This step opens the way to interrogate the cell's spatio-temporal evolution with molecular dynamics simulations, an approach that can be extended to other cell types in the near future.

Computing the Volume, Surface Area, Mean, and Gaussian Curvatures of Molecules and Their Derivatives

Geometry is crucial in our efforts to comprehend the structures and dynamics of biomolecules. For example, volume, surface area, and integrated mean and Gaussian curvature of the union of balls representing a molecule are used to quantify its interactions with the water surrounding it in the morphometric implicit solvent models. The Alpha Shape theory provides an accurate and reliable method for computing these geometric measures. In this paper, we derive homogeneous formulas for the expressions of these measures and their derivatives with respect to the atomic coordinates, and we provide algorithms that implement them into a new software package, AlphaMol. The only variables in these formulas are the interatomic distances, making them insensitive to translations and rotations. AlphaMol includes a sequential algorithm and a parallel algorithm. In the parallel version, we partition the atoms of the molecule of interest into 3D rectangular blocks, using a kd-tree algorithm. We then apply the sequential algorithm of AlphaMol to each block, augmented by a buffer zone to account for atoms whose ball representations may partially cover the block. The current parallel version of AlphaMol leads to a 20-fold speed-up compared to an independent serial implementation when using 32 processors. For instance, it takes 31 s to compute the geometric measures and derivatives of each atom in a viral capsid with more than 26 million atoms on 32 Intel processors running at 2.7 GHz. The presence of the buffer zones, however, leads to redundant computations, which ultimately limit the impact of using multiple processors. AlphaMol is available as an OpenSource software.

LongBondEliminator: A Molecular Simulation Tool to Remove Ring Penetrations in Biomolecular Simulation Systems

We develop a workflow, implemented as a plugin to the molecular visualization program VMD, that can fix ring penetrations with minimal user input. LongBondEliminator, detects ring piercing artifacts by the long, strained bonds that are the local minimum energy conformation during minimization for some assembled simulation system. The LongBondEliminator tool then automatically treats regions near these long bonds using multiple biases applied through NAMD. By combining biases implemented through the collective variables module, density-based forces, and alchemical techniques in NAMD, LongBondEliminator will iteratively alleviate long bonds found within molecular simulation systems. Through three concrete examples with increasing complexity, a lignin polymer, an viral capsid assembly, and a large, highly glycosylated protein aggrecan, we demonstrate the utility for this method in eliminating ring penetrations from classical MD simulation systems. The tool is available via gitlab as a VMD plugin, and has been developed to be generically useful across a variety of biomolecular simulations.

A brief history of visualizing membrane systems in molecular dynamics simulations

Understanding lipid dynamics and function, from the level of single, isolated molecules to large assemblies, is more than ever an intensive area of research. The interactions of lipids with other molecules, particularly membrane proteins, are now extensively studied. With advances in the development of force fields for molecular dynamics simulations (MD) and increases in computational resources, the creation of realistic and complex membrane systems is now common. In this perspective, we will review four decades of the history of molecular dynamics simulations applied to membranes and lipids through the prism of molecular graphics.

Mechanisms of E. coli chemotaxis signaling pathways visualized using cryoET and computational approaches

Chemotaxis signaling pathways enable bacteria to sense and respond to their chemical environment and, in some species, are critical for lifestyle processes such as biofilm formation and pathogenesis. The signal transduction underlying chemotaxis behavior is mediated by large, highly ordered protein complexes known as chemosensory arrays. For nearly two decades, cryo-electron tomography (cryoET) has been used to image chemosensory arrays, providing an increasingly detailed understanding of their structure and function. In this mini-review, we provide an overview of the use of cryoET to study chemosensory arrays, including imaging strategies, key results, and outstanding questions. We further discuss the application of molecular modeling and simulation techniques to complement structure determination efforts and provide insight into signaling mechanisms. We close the review with a brief outlook, highlighting promising future directions for the field.

MDsrv: visual sharing and analysis of molecular dynamics simulations

Molecular dynamics simulation is a proven technique for computing and visualizing the time-resolved motion of macromolecules at atomic resolution. The MDsrv is a tool that streams MD trajectories and displays them interactively in web browsers without requiring advanced skills, facilitating interactive exploration and collaborative visual analysis. We have now enhanced the MDsrv to further simplify the upload and sharing of MD trajectories and improve their online viewing and analysis. With the new instance, the MDsrv simplifies the creation of sessions, which allows the exchange of MD trajectories with preset representations and perspectives. An important innovation is that the MDsrv can now access and visualize trajectories from remote datasets, which greatly expands its applicability and use, as the data no longer needs to be accessible on a local server. In addition, initial analyses such as sequence or structure alignments, distance measurements, or RMSD calculations have been implemented, which optionally support visual analysis. Finally, based on Mol*, MDsrv now provides faster and more efficient visualization of even large trajectories compared to its predecessor tool NGL.