Exploiting Sequence-Dependent Rotamer Information in Global Optimization of Proteins.

Rotamers, namely amino acid side chain conformations common to many different peptides, can be compiled into libraries. These rotamer libraries are used in protein modeling, where the limited conformational space occupied by amino acid side chains is exploited. Here, we construct a sequence-dependent rotamer library from simulations of all possible tripeptides, which provides rotameric states dependent on adjacent amino acids. We observe significant sensitivity of rotamer populations to sequence and find that the library is successful in locating side chain conformations present in crystal structures. The library is designed for applications with basin-hopping global optimization, where we use it to propose moves in conformational space. The addition of rotamer moves significantly increases the efficiency of protein structure prediction within this framework, and we determine parameters to optimize efficiency.

Elucidating the solution structure of the K-means cost function using energy landscape theory.

The K-means algorithm, routinely used in many scientific fields, generates clustering solutions that depend on the initial cluster coordinates. The number of solutions may be large, which can make locating the global minimum challenging. Hence, the topography of the cost function surface is crucial to understanding the performance of the algorithm. Here, we employ the energy landscape approach to elucidate the topography of the K-means cost function surface for Fisher's Iris dataset. For any number of clusters, we find that the solution landscapes have a funneled structure that is usually associated with efficient global optimization. An analysis of the barriers between clustering solutions shows that the funneled structures result from remarkably small barriers between almost all clustering solutions. The funneled structure becomes less well-defined as the number of clusters increases, and we analyze kinetic analogs to quantify the increased difficulty in locating the global minimum for these different landscapes.

Improving double-ended transition state searches for soft-matter systems.

Transitions between different stable configurations of biomolecules are important in understanding disease mechanisms, structure-function relations, and novel molecular-scale engineering. The corresponding pathways can be characterized efficiently using geometry optimization schemes based on double-ended transition state searches. An interpolation is first constructed between the known states and then refined, yielding a band that contains transition state candidates. Here, we analyze an example where various interpolation schemes lead to bands with a single step transition, but the correct pathway actually proceeds via an intervening, low-energy minimum. We compare a number of different interpolation schemes for this problem. We systematically alter the number of discrete images in the interpolations and the spring constants used in the optimization and test two schemes for adjusting the spring constants and image distribution, resulting in a total of 2760 different connection attempts. Our results confirm that optimized bands are not necessarily a good description of the transition pathways in themselves, and further refinement to actually converge transition states and establish their connectivity is required. We see an improvement in the optimized bands if we employ the adjustment of spring constants with doubly-nudged elastic band and a smaller improvement from the image redistribution. The example we consider is representative of numerous cases we have encountered in a wide variety of molecular and condensed matter systems.

Tunneling Splittings in Water Clusters from Path Integral Molecular Dynamics.

We present calculations of tunneling splittings in selected small water clusters, based on a recently developed path integral molecular dynamics (PIMD) method. The ground-rotational-state tunneling motions associated with the largest splittings in the water dimer, trimer, and hexamer are considered, and we show that the PIMD predictions are in very good agreement with benchmark quantum and experimental results. As the tunneling spectra are highly sensitive to both the details of the quantum dynamics and the potential energy surface, our calculations are a validation of the MB-Pol surface as well as the accuracy of PIMD. The favorable scaling of PIMD with system size paves the way for calculations of tunneling splittings in large, nonrigid molecular systems with motions that cannot be treated accurately by other methods, such as the semiclassical instanton.

Morphological analysis of chiral rod clusters from a coarse-grained single-site chiral potential.

We present a coarse-grained single-site potential for simulating chiral interactions, with adjustable strength, handedness, and preferred twist angle. As an application, we perform basin-hopping global optimisation to predict the favoured geometries for clusters of chiral rods. The morphology phase diagram based upon these predictions has four distinct families, including previously reported structures for potentials that introduce chirality based on shape, such as membranes and helices. The transition between these two configurations reproduces some key features of experimental results for fd bacteriophage. The potential is computationally inexpensive, intuitive, and versatile; we expect it will be useful for large scale simulations of chiral molecules. For chiral particles confined in a cylindrical container we reproduce the behaviour observed for fusilli pasta in a jar. Hence this chiropole potential has the capability to provide insight into structures on both macroscopic and molecular length scales.

Effects of random pinning on the potential energy landscape of a supercooled liquid.

We use energy landscape methods to investigate the response of a supercooled liquid to random pinning. We classify the structural similarity of different energy minima using a measure of overlap. This analysis reveals a correspondence between distinct particle packings (which are characterised via the overlap) and funnels on the energy landscape (which are characterised via disconnectivity graphs). As the number of pinned particles is increased, we find a crossover from glassy behavior at low pinning to a structure-seeking landscape at high pinning, in which all thermally accessible minima are structurally similar. We discuss the consequences of these results for theories of randomly pinned liquids. We also investigate how the energy landscape depends on the fraction of pinned particles, including the degree of frustration and the evolution of distinct packings as the number of pinned particles is reduced.

Tunneling splittings from path-integral molecular dynamics using a Langevin thermostat.

We report an improved method for the calculation of tunneling splittings between degenerate configurations in molecules and clusters using path-integral molecular dynamics (PIMD). Starting from an expression involving a ratio of thermodynamic density matrices at the bottom of the symmetric wells, we use thermodynamic integration with molecular dynamics simulations and a Langevin thermostat to compute the splittings stochastically. The thermodynamic integration is performed by sampling along the semiclassical instanton path, which provides an efficient reaction coordinate as well as being physically well-motivated. This approach allows us to carry out PIMD calculations of the multi-well tunneling splitting pattern in the water dimer and to refine previous PIMD calculations for one-dimensional models and malonaldehyde. The large (acceptor) splitting in the water dimer agrees to within 20% of benchmark variational results, and the smaller splittings agree to within 10%.

Structure, thermodynamics, and rearrangement mechanisms in gold clusters-insights from the energy landscapes framework.

We consider finite-size and temperature effects on the structure of model AuN clusters (30 ≤ N ≤ 147) bound by the Gupta potential. Equilibrium behaviour is examined in the harmonic superposition approximation, and the size-dependent melting temperature is also bracketed using molecular dynamics simulations. We identify structural transitions between distinctly different morphologies, characterised by various defect features. Reentrant behaviour and trends with respect to cluster size and temperature are discussed in detail. For N = 55, 85, and 147 we visualise the topography of the underlying potential energy landscape using disconnectivity graphs, colour-coded by the cluster morphology; and we use discrete path sampling to characterise the rearrangement mechanisms between competing structures separated by high energy barriers (up to 1 eV). The fastest transition pathways generally involve metastable states with multiple fivefold disclinations and/or a high degree of amorphisation, indicative of melting. For N = 55 we find that reoptimising low-lying minima using density functional theory (DFT) alters their energetic ordering and produces a new putative global minimum at the DFT level; however, the equilibrium structure predicted by the Gupta potential at room temperature is consistent with previous experiments.

Pathways for diffusion in the potential energy landscape of the network glass former SiO2.

We study the dynamical behaviour of a computer model for viscous silica, the archetypal strong glass former, and compare its diffusion mechanism with earlier studies of a fragile binary Lennard-Jones liquid. Three different methods of analysis are employed. First, the temperature and time scale dependence of the diffusion constant is analysed. Negative correlation of particle displacements influences transport properties in silica as well as in fragile liquids. We suggest that the difference between Arrhenius and super-Arrhenius diffusive behaviour results from competition between the correlation time scale and the caging time scale. Second, we analyse the dynamics using a geometrical definition of cage-breaking transitions that was proposed previously for fragile glass formers. We find that this definition accurately captures the bond rearrangement mechanisms that control transport in open network liquids, and reproduces the diffusion constants accurately at low temperatures. As the same method is applicable to both strong and fragile glass formers, we can compare correlation time scales in these two types of systems. We compare the time spent in chains of correlated cage breaks with the characteristic caging time and find that correlations in the fragile binary Lennard-Jones system persist for an order of magnitude longer than those in the strong silica system. We investigate the origin of the correlation behaviour by sampling the potential energy landscape for silica and comparing it with the binary Lennard-Jones model. We find no qualitative difference between the landscapes, but several metrics suggest that the landscape of the fragile liquid is rougher and more frustrated. Metabasins in silica are smaller than those in binary Lennard-Jones and contain fewer high-barrier processes. This difference probably leads to the observed separation of correlation and caging time scales.

Structure and Thermodynamics of Metal Clusters on Atomically Smooth Substrates.

We analyze the structure of model NiN and CuN clusters (N = 55, 147) supported on a variety of atomically smooth van der Waals surfaces. The global minima are mapped in the space of two parameters: (i) the laterally averaged surface stickiness, γ, which controls the macroscopic wetting angle, and (ii) the surface microstructure, which produces more subtle but important templating via epitaxial stresses. We find that adjusting the substrate lattice (even at constant γ) can favor different crystal plane orientations in the cluster, stabilize hexagonal close-packed order, or induce various defects, such as stacking faults, twin boundaries, and five-fold disclinations. Thermodynamic analysis reveals substrate-dependent solid-solid transitions in cluster morphology, with tunable transition temperature and sometimes exhibiting re-entrant behavior. These results shed new light on the extent to which a supporting surface can be used to influence the equilibrium behavior of nanoparticles.

Defining and quantifying frustration in the energy landscape: Applications to atomic and molecular clusters, biomolecules, jammed and glassy systems.

The emergence of observable properties from the organisation of the underlying potential energy landscape is analysed, spanning a full range of complexity from self-organising to glassy and jammed systems. The examples include atomic and molecular clusters, a ß-barrel protein, the GNNQQNY peptide dimer, and models of condensed matter that exhibit structural glass formation and jamming. We have considered measures based on several different properties, namely, the Shannon entropy, an equilibrium thermodynamic measure that uses a sample of local minima, and indices that require additional information about the connections between local minima in the form of transition states. A frustration index is defined that correlates directly with key properties that distinguish relaxation behaviour within this diverse set. The index uses the ratio of the energy barrier to the energy difference with reference to the global minimum. The contributions for each local minimum are weighted by the equilibrium occupation probabilities. Hence we obtain fundamental insight into the connections and distinctions between systems that cover the continuum from efficient structure-seekers to landscapes that exhibit broken ergodicity and rare event dynamics.

Dynamics and thermodynamics of the coronene octamer described by coarse-grained potentials.

Coarse-grained models developed for polycyclic aromatic hydrocarbons based on the Paramonov-Yaliraki potential have been employed to investigate the finite temperature thermodynamics, out-of-equilibrium dynamics, energy landscapes, and rearrangement pathways of the coronene octamer. Molecular dynamics simulations are used to address the short-time behaviour, diffusion properties, convergence to equilibrium, and dissociation kinetics. A kinetic transition network composed of a connected database of stationary points provides a consistent picture of the complex potential and free energy landscapes, and enables us to describe rearrangements occurring over long time scales and associated thermal properties. Comparison with reference simulations performed with an all-atom description, indicates satisfactory agreement at moderate energies, especially when quadrupole corrections to the intermolecular potential are included. At higher energies, unimolecular evaporation rates are particularly well reproduced by the coarse-grained model. The potential energy landscapes exhibit multiple funnels for all the models, with alternative columnar arrangements competing at low energy. Entropy-driven structural transitions are predicted to involve largely cooperative motion, with entire stacks shifting and rotating around one another. These structural transitions, which were not characterised in earlier parallel tempering Monte Carlo simulations, are well represented by the coarse-grained models, with similar barrier heights but fewer steps.

Impurity effects on solid-solid transitions in atomic clusters.

We use the harmonic superposition approach to examine how a single atom substitution affects low-temperature anomalies in the vibrational heat capacity (CV) of model nanoclusters. Each anomaly is linked to competing solidlike "phases", where crossover of the corresponding free energies defines a solid-solid transition temperature (Ts). For selected Lennard-Jones clusters we show that Ts and the corresponding CV peak can be tuned over a wide range by varying the relative atomic size and binding strength of the impurity, but excessive atom-size mismatch can destroy a transition and may produce another. In some tunable cases we find up to two additional CV peaks emerging below Ts, signalling one- or two-step delocalisation of the impurity within the ground-state geometry. Results for Ni74X and Au54X clusters (X = Au, Ag, Al, Cu, Ni, Pd, Pt, Pb), modelled by the many-body Gupta potential, further corroborate the possibility of tuning, engineering, and suppressing finite-system analogues of a solid-solid transition in nanoalloys.

Self-assembly of colloidal magnetic particles: energy landscapes and structural transitions.

The self-assembly of colloidal magnetic particles is of particular interest for the rich variety of structures it produces and the potential for these systems to be reconfigurable. In the present study we characterised the structures for clusters of N spherical colloidal magnetic particles in the presence of short-ranged attractive depletion interactions up to N = 50. The morphologies that we observed include linear chains, circular rings, stacks of two and three circular rings, as well as compact structures consisting of sheets. For size-selected clusters we illustrate the organisation of the low-lying part of the potential energy landscape, and analyse pathways for the structural transitions of interest, including the effect of an external static magnetic field.

Dynamics of a molecular glass former: Energy landscapes for diffusion in ortho-terphenyl.

Relaxation times and transport processes of many glass-forming supercooled liquids exhibit a super-Arrhenius temperature dependence. We examine this phenomenon by computer simulation of the Lewis-Wahnström model for ortho-terphenyl. We propose a microscopic definition for a single-molecule cage-breaking transition and show that, when correlation behaviour is taken into account, these rearrangements are sufficient to reproduce the correct translational diffusion constants over an intermediate temperature range in the supercooled regime. We show that super-Arrhenius behaviour can be attributed to increasing negative correlation in particle movement at lower temperatures and relate this to the cage-breaking description. Finally, we sample the potential energy landscape of the model and show that it displays hierarchical ordering. Substructures in the landscape, which may correspond to metabasins, have boundaries defined by cage-breaking transitions. The cage-breaking formulation provides a direct link between the potential energy landscape and macroscopic diffusion behaviour.

Coarse-graining the structure of polycyclic aromatic hydrocarbons clusters.

Clusters of polycyclic aromatic hydrocarbons (PAHs) are essential components of soot and may concentrate a significant fraction of carbon matter in the interstellar medium. In this contribution, coarse-grained potentials are parameterized using all-atom reference data to model PAH molecules, such as coronene (C24H12) or circumcoronene (C54H18), and their aggregates. Low-energy structures of pure coronene or circumcoronene clusters obtained using basin-hopping global optimization are found to agree with atomistic results, and consist of finite 1D columnar motifs, sometimes juxtaposed in larger clusters. The structures are only weakly perturbed when quadrupolar interactions are included. π-Stacking also dominates in binary coronene/circumcoronene aggregates, although intriguing motifs are predicted in which one or more molecules are sandwiched between the other PAH species. The coarse-grained model is also extended to account for interaction with a flat graphitic substrate. In this case, binding is stronger with the substrate than with other molecules, and the PAHs are predicted to arrange into a flat triangular monolayer.

Dynamical properties of two- and three-dimensional colloidal clusters of six particles.

Colloidal clusters are important systems for studying self-assembly. Clusters of six colloidal particles attracting each other via short-ranged interactions have been recently studied both theoretically and experimentally. Here we present a computer modelling study of the thermodynamics and dynamics of these clusters using a short-ranged Morse potential in two and three dimensions. We combine energy landscape methods with comprehensive sampling, both of configurations using Markov chain Monte Carlo and also of trajectories using Langevin molecular dynamics propagation. We show that the interaction energies between the particles are probably greater than previously assumed. The rates predicted by transition state theory using harmonic vibrational densities of states are off by four orders of magnitude, since the effects of viscosity are not accounted for. In contrast, sampling short trajectories using an appropriate friction constant and discrete relaxation path sampling produces reasonable agreement with the experimental rates.

Grand and Semigrand Canonical Basin-Hopping.

We introduce grand and semigrand canonical global optimization approaches using basin-hopping with an acceptance criterion based on the local contribution of each potential energy minimum to the (semi)grand potential. The method is tested using local harmonic vibrational densities of states for atomic clusters as a function of temperature and chemical potential. The predicted global minima switch from dissociated states to clusters for larger values of the chemical potential and lower temperatures, in agreement with the predictions of a model fitted to heat capacity data for selected clusters. Semigrand canonical optimization allows us to identify particularly stable compositions in multicomponent nanoalloys as a function of increasing temperature, whereas the grand canonical potential can produce a useful survey of favorable structures as a byproduct of the global optimization search.

Markov state modeling and dynamical coarse-graining via discrete relaxation path sampling.

A method is derived to coarse-grain the dynamics of complex molecular systems to a Markov jump process (MJP) describing how the system jumps between cells that fully partition its state space. The main inputs are relaxation times for each pair of cells, which are shown to be robust with respect to positioning of the cell boundaries. These relaxation times can be calculated via molecular dynamics simulations performed in each cell separately and are used in an efficient estimator for the rate matrix of the MJP. The method is illustrated through applications to Sinai billiards and a cluster of Lennard-Jones discs.

Quasi-combinatorial energy landscapes for nanoalloy structure optimisation.

We formulate nanoalloy structure prediction as a mixed-variable optimisation problem, where the homotops can be associated with an effective, quasi-combinatorial energy landscape in permutation space. We survey this effective landscape for a representative set of binary systems modelled by the Gupta potential. In segregating systems with small lattice mismatch, we find that homotops have a relatively straightforward landscape with few local optima - a scenario well-suited for local (combinatorial) optimisation techniques that scale quadratically with system size. Combining these techniques with multiple local-neighbourhood structures yields a search for multiminima, and we demonstrate that generalised basin-hopping with a metropolis acceptance criterion in the space of multiminima can then be effective for global optimisation of binary and ternary nanoalloys.