Scaling of Multimillion-Atom Biological Molecular Dynamics Simulation on a Petascale Supercomputer

Biomolecular dynamics: order-disorder transitions and energy landscapes

While the energy landscape theory of protein folding is now a widely accepted view for understanding how relatively weak molecular interactions lead to rapid and cooperative protein folding, such a framework must be extended to describe the large-scale functional motions observed in molecular machines. In this review, we discuss (1) the development of the energy landscape theory of biomolecular folding, (2) recent advances toward establishing a consistent understanding of folding and function and (3) emerging themes in the functional motions of enzymes, biomolecular motors and other biomolecular machines. Recent theoretical, computational and experimental lines of investigation have provided a very dynamic picture of biomolecular motion. In contrast to earlier ideas, where molecular machines were thought to function similarly to macroscopic machines, with rigid components that move along a few degrees of freedom in a deterministic fashion, biomolecular complexes are only marginally stable. Since the stabilizing contribution of each atomic interaction is on the order of the thermal fluctuations in solution, the rigid body description of molecular function must be revisited. An emerging theme is that functional motions encompass order-disorder transitions and structural flexibility provides significant contributions to the free energy. In this review, we describe the biological importance of order-disorder transitions and discuss the statistical-mechanical foundation of theoretical approaches that can characterize such transitions.

Computer Simulations of Voltage-Gated Cation Channels

The relentless growth in computational power has seen increasing applications of molecular dynamics (MD) simulation to the study of membrane proteins in realistic membrane environments, which include explicit membrane lipids, water and ions. The concomitant increasing availability of membrane protein structures for ion channels, and transporters -- to name just two examples -- has stimulated many of these MD studies. In the case of voltage-gated cation channels (VGCCs) recent computational works have focused on ion-conduction and gating mechanisms, along with their regulation by agonist/antagonist ligands. The information garnered from these computational studies is largely inaccessible to experiment and is crucial for understanding the interplay between the structure and function as well as providing new directions for experiments. This article highlights recent advances in probing the structure and function of potassium channels and offers a perspective on the challenges likely to arise in making analogous progress in characterizing sodium channels.

Molecular dynamics methods for modeling complex interactions in biomaterials

The molecular dynamics method is a powerful computer simulation technique which provides access to the detailed time evolution (trajectory) of a system in specified conditions, such as a particular temperature or pressure. The full trajectory of the system can be analyzed using statistical mechanics tools to obtain thermodynamical quantities and dynamical properties; the mechanism of chemical reactions and other time-dependent processes, such as diffusion, can also be revealed in high detail. When applied to model extended and complex system such as biomaterials, MD simulations represent an invaluable tool to discover structure-activity relationships and rationalize biomedical applications.

Nanometrology of delignified Populus using mode synthesizing atomic force microscopy

The study of the spatially resolved physical and compositional properties of materials at the nanoscale is increasingly challenging due to the level of complexity of biological specimens such as those of interest in bioenergy production. Mode synthesizing atomic force microscopy (MSAFM) has emerged as a promising metrology tool for such studies. It is shown that, by tuning the mechanical excitation of the probe-sample system, MSAFM can be used to dynamically investigate the multifaceted complexity of plant cells. The results are argued to be of importance both for the characteristics of the invoked synthesized modes and for accessing new features of the samples. As a specific system to investigate, we present images of Populus, before and after a holopulping treatment, a crucial step in the biomass delignification process.

Parallel Generalized Born Implicit Solvent Calculations with NAMD

Accurate electrostatic descriptions of aqueous solvent are critical for simulation studies of bio-molecules, but the computational cost of explicit treatment of solvent is very high. A computationally more feasible alternative is a generalized Born implicit solvent description which models polar solvent as a dielectric continuum. Unfortunately, the attainable simulation speedup does not transfer to the massive parallel computers often employed for simulation of large structures. Longer cutoff distances, spatially heterogenous distribution of atoms and the necessary three-fold iteration over atom-pairs in each timestep combine to challenge efficient parallel performance of generalized Born implicit solvent algorithms. Here we report how NAMD, a parallel molecular dynamics program, meets the challenge through a unique parallelization strategy. NAMD now permits efficient simulation of large systems whose slow conformational motions benefit most from implicit solvent descriptions due to the inherent low viscosity. NAMD's implicit solvent performance is benchmarked and then illustrated in simulating the ratcheting Escherichia coli ribosome involving ~250,000 atoms.

Multiscale simulation of microbe structure and dynamics

A multiscale mathematical and computational approach is developed that captures the hierarchical organization of a microbe. It is found that a natural perspective for understanding a microbe is in terms of a hierarchy of variables at various levels of resolution. This hierarchy starts with the N -atom description and terminates with order parameters characterizing a whole microbe. This conceptual framework is used to guide the analysis of the Liouville equation for the probability density of the positions and momenta of the N atoms constituting the microbe and its environment. Using multiscale mathematical techniques, we derive equations for the co-evolution of the order parameters and the probability density of the N-atom state. This approach yields a rigorous way to transfer information between variables on different space-time scales. It elucidates the interplay between equilibrium and far-from-equilibrium processes underlying microbial behavior. It also provides framework for using coarse-grained nanocharacterization data to guide microbial simulation. It enables a methodical search for free-energy minimizing structures, many of which are typically supported by the set of macromolecules and membranes constituting a given microbe. This suite of capabilities provides a natural framework for arriving at a fundamental understanding of microbial behavior, the analysis of nanocharacterization data, and the computer-aided design of nanostructures for biotechnical and medical purposes. Selected features of the methodology are demonstrated using our multiscale bionanosystem simulator DeductiveMultiscaleSimulator. Systems used to demonstrate the approach are structural transitions in the cowpea chlorotic mosaic virus, RNA of satellite tobacco mosaic virus, virus-like particles related to human papillomavirus, and iron-binding protein lactoferrin.

A Solvent-Free Coarse Grain Model for Crystalline and Amorphous Cellulose Fibrils

Understanding biomass structure and dynamics on a range of time and length scales is important for the development of cellulosic biofuels. Here, to enable length and time scale extension, we develop a coarse grain (CG) model for molecular dynamics (MD) simulations of cellulose. For this purpose, we use distribution functions from fully atomistic MD simulations as target observables. A single bead per monomer level coarse graining is found to be sufficient to successfully reproduce structural features of crystalline cellulose. Without the use of constraints the CG crystalline fibril is found to remain stable over the maximum simulation length explored in this study (>1 μs). We also extend the CG representation to model fully amorphous cellulose fibrils. This is done by using an atomistic MD simulation of fully solvated individual cellulose chains as a target for developing the corresponding fully amorphous CG force field. Fibril structures with different degrees of crystallinity are obtained using force fields derived using a parameter coupling the crystalline and amorphous potentials. The method provides an accurate and constraint-free approach to derive CG models for cellulose with a wide range of crystallinity, suitable for incorporation into large-scale models of lignocellulosic biomass.

Self-similar multiscale structure of lignin revealed by neutron scattering and molecular dynamics simulation

Lignin, a major polymeric component of plant cell walls, forms aggregates in vivo and poses a barrier to cellulosic ethanol production. Here, neutron scattering experiments and molecular dynamics simulations reveal that lignin aggregates are characterized by a surface fractal dimension that is invariant under change of scale from ~1-1000 Å. The simulations also reveal extensive water penetration of the aggregates and heterogeneous chain dynamics corresponding to a rigid core with a fluid surface.

Effect of atom- and group-based truncations on biomolecules simulated with reaction-field electrostatics

The performance of the reaction-field method of electrostatics is tested in molecular dynamics simulations of protein human interleukin-4 and a short DNA fragment in explicit solvent. Two truncation schemes are considered: one based on the position of atomic charges in water molecules and the other on the position of groups of charges. The group-based truncation leads to the melting of the DNA double helix. In contrast, the atom-based truncation maintains the helical structure intact. Similarly for the protein, the group-based truncation leads to an unfolding at pH 2 while the atom-based truncation produces stable trajectories at low and normal pH, in agreement with experiment. Artificial repulsion between charged residues associated with the group-based truncation is identified as the microscopic reason behind unfolding of the protein. Implications of different truncation schemes in reaction-field simulations of biomolecules are discussed.

Trajectory NG: portable, compressed, general molecular dynamics trajectories

We present general algorithms for the compression of molecular dynamics trajectories. The standard ways to store MD trajectories as text or as raw binary floating point numbers result in very large files when efficient simulation programs are used on supercomputers. Our algorithms are based on the observation that differences in atomic coordinates/velocities, in either time or space, are generally smaller than the absolute values of the coordinates/velocities. Also, it is often possible to store values at a lower precision. We apply several compression schemes to compress the resulting differences further. The most efficient algorithms developed here use a block sorting algorithm in combination with Huffman coding. Depending on the frequency of storage of frames in the trajectory, either space, time, or combinations of space and time differences are usually the most efficient. We compare the efficiency of our algorithms with each other and with other algorithms present in the literature for various systems: liquid argon, water, a virus capsid solvated in 15 mM aqueous NaCl, and solid magnesium oxide. We perform tests to determine how much precision is necessary to obtain accurate structural and dynamic properties, as well as benchmark a parallelized implementation of the algorithms. We obtain compression ratios (compared to single precision floating point) of 1:3.3-1:35 depending on the frequency of storage of frames and the system studied.

Efficient solutions of self-consistent mean field equations for dewetting and electrostatics in nonuniform liquids

We use a new configuration-based version of linear response theory to efficiently solve self-consistent mean field equations relating an effective single particle potential to the induced density. The versatility and accuracy of the method is illustrated by applications to dewetting of a hard sphere solute in a Lennard-Jones fluid, the interplay between local hydrogen bond structure and electrostatics for water confined between two hydrophobic walls, and to ion pairing in ionic solutions. Simulation time has been reduced by more than an order of magnitude over previous methods.

Importance of the lactate dehydrogenase quaternary structure in theoretical calculations

Using the example of lactate dehydrogenase, we show that enzyme quaternary structure has an important influence on the structure of the active site and that models that comprise all amino acids in the vicinity of an active site, but are missing this structural information, can lead to incorrect results. We also show that binding isotope effects are very sensitive to the geometric parameters, and thus one should be very cautious when interpreting results obtained with models that are too coarse. In terms of the type of hydrogen bonds, our results indicate that binding isotope effects are pronounced only when a hydrogen bond exhibits some covalent character.

A hierarchic collision detection algorithm for simple Brownian dynamics

We describe an algorithm to avoid steric violation (bumps) between bodies arranged in a hierarchy. The algorithm recursively directs the focus of a bump-detector towards the interactions of children whose parents are in collision. This has the effect of concentrating available computer resources towards maintaining good steric interactions in the region where bodies are colliding. The algorithm was implemented and tested under two programming environments: a graphical environment, OpenGL under Java3D, and a non-graphical environment in "C". The former used a built-in collision detection system whereas the latter used a simple algorithm devised previously for the interaction of "soft" bodies. This simpler system was found to run much faster (by 50-fold) even after allowing for time spent on graphical activity and was also better at preventing steric violations. With a hierarchy of three levels of 100, the non-graphical implementation was able to simulate a million atomic bodies for 100,000 steps in 12h on a laptop computer.