Molecular Dynamics Day


"Molecular Dynamics Day"

Monday 13th June 2016, 1.00 pm – 6.00 pm
Department of Statistics, 24-29 St Giles’, Oxford OX1 3LB

All are welcome. To assist with catering, please register your attendance via email to Beverley Lane ( with your name, affiliation and any special dietary requirements


1.00 pm    
Jotun Hein (Oxford) - Welcome
1.10 pm
Ben Leimkuhler (Edinburgh)

Title: Stochastic-numerical algorithms for fast sampling in high dimensions

Abstract: Molecular models and statistical inference problems give rise to gargantuan systems of stochastic differential equations (SDEs) whose paths ergodically sample multimodal probability distributions.  An important challenge for the numerical analyst (or the chemist, or the physicist, or the engineer, or the statistician) is the design of efficient numerical methods to generate these paths.   For SDEs, the numerical perspective is just maturing, with important new methods (and, even more important, new procedures for their construction and analysis) becoming available [1]. One of the interesting ideas is to design stochastic schemes with close attention to the error in invariant measures.  Another is to use negative feedback loop controls to regulate a noisy gradient or even the discretisation error itself.  To illustrate our approach, I will discuss several different examples including (i) efficient schemes for constrained stochastic dynamics improving accuracy and stability in bio-MD [2],  (ii) methods for  Bayesian sampling for machine learning applications [3] and (iii) pairwise adaptive thermostats for nonequilibrium mesoscale simulation (an improved "dissipative particle dynamics" method) [4].   [1] B. Leimkuhler and C. Matthews, Molecular Dynamics, Springer 2015. [2] B. Leimkuhler and C. Matthews, Efficient molecular dynamics using geodesic integration and solvent-solute splitting, Proc. Roy. Soc. A, in press, 2016. [3] X. Shang, Z. Zhu, B. Leimkuhler and A. Storkey, Covariance-controlled adaptive Langevin thermostat for large-scale Bayesian sampling, NIPS 2015. [4] B. Leimkuhler and X. Shang, Pairwise adaptive thermostats for improved accuracy and stability in dissipative particle dynamics, preprint, 2016.

2.10 pm
David Manolopoulos (Oxford)

Title: Ring polymer molecular dynamics

Abstract: Quantum mechanical zero point energy and tunnelling effects in the nuclear dynamics play a significant role in many systems containing hydrogen atoms, but these effects are typically ignored in molecular dynamics simulations. Ring polymer molecular dynamics (RPMD) provides a practical way to include them, at least approximately. I will spend this 25 minutes explaining how.

2.35 pm
Anna Duncan (University of Oxford)

Title: Complex and crowded membranes under the computational microscope

Abstract: It is well-established that cell membranes are complex, crowded environments.  However, it is difficult to obtain experimental data about these systems that is both high resolution and includes dynamic behaviour and in vivo complexity. Using relatively large scale molecular dynamics simulations, we can observe membrane behaviour, such as the interplay between protein and lipid diffusion, protein-lipid interactions and protein clustering. Here I will present two cases: the effect of lipid composition on clustering of potassium channels and clustering of proteins in the E. coli outer membrane.  I will also talk about how we can best compare our simulation results with experimental data.)

3.00 pm
Coffee and Discussion
3.20 pm
Maria Musgaard (Oxford)

Title: Steered Molecular Dynamics Simulations Predict Conformational Stability of Glutamate Receptors

Abstract: The stability of protein-protein interfaces can be essential for protein function. For ionotropic glutamate receptors, a family of ligand-gated ion channels vital for normal function of the central nervous system, such an interface exists between the extracellular ligand binding domains (LBDs). In the full-length protein, the LBDs are arranged as a dimer of dimers. Agonist binding to the LBDs opens the ion channel, and briefly after activation the receptor desensitizes. Several residues at the LBD dimer interface are known to modulate desensitization and conformational changes around these residues are believed to be involved in the state transition. The general hypothesis is that the interface is disrupted upon desensitization, and structural evidence suggests that the disruption might be substantial. However, when crosslinking the central part of this interface, functional data suggest that the receptor can still undergo desensitization, contradicting the hypothesis of major interface disruption.

Here, we illustrate how opening the dimer interface using steered molecular dynamics (SMD) simulations, and analyzing the work values required, provides a quantitative measure for interface stability.  For one subtype of glutamate receptors, which is regulated by ion binding to the dimer interface, we show that opening the interface without ions bound requires less work than with ions present, suggesting that ion binding indeed stabilizes the interface. Likewise, for interface mutants with longer-lived active states, the interface is more stable, while the work required to open the interface is reduced for less active mutants. Moreover, a crosslinked mutant can still undergo initial interface opening motions similar to the native receptor and at similar energetic cost. Thus, our results support that interface opening is involved in desensitization. Furthermore, they provide reconciliation of apparently opposing data and demonstrate that SMD simulations can give relevant biological insight into longer timescale processes without the need for expensive calculations.

3.45 pm
Samuel Livingstone [Bristol] 

Title: On the geometric ergodicity of Hybrid Monte Carlo

Abstract: Hamiltonian/Hybrid Monte Carlo (HMC) is a sampling method which has existed for almost 30 years, and recently has become very popular among statisticians, primarily because general purpose software for its implementation is now available.  In this talk I'll discuss recent work in which we establish fairly general \pi-irreducibility and geometric ergodicity criteria for the method, giving some basic guidelines on when it should 'work well' for estimating expectations of interest.  The results also shed light on how to tune some of the free parameters.  If time permits I may also mention some ongoing work on non-quadratic choices for the kinetic energy, and how this can either positively or negatively impact on geometric ergodicity.

4.10 pm
Dr. Michael Betancourt [Warwick]

Title: Adiabatic Monte Carlo

Abstract: By using local information to guide the exploration of a target distribution, Markov Chain Monte Carlo, in particular modern implementations like Hamiltonian Monte Carlo, has been a cornerstone of modern statistical computation.  Unfortunately this local information is not generally sufficient to admit computations that require global information, such as estimating expectations with respect to multimodal distributions or marginal likelihoods.  When coupled with an interpolation between the target distribution and a simpler auxiliary distribution, however, Markov Chain Monte Carlo can be an important component, for example in simulated annealing, simulated tempering, and their variants.  Unfortunately, determining an effective interpolation is a challenging tuning problem that hampers these methods in practice.

In this talk I will show how the same differential geometry from which Hamiltonian Monte Carlo is built can also be used to construct an optimal interpolation dynamically, with no user intervention.  I will then present the resulting Adiabatic Monte Carlo algorithm with discussion of its promise and some of the open problems in its general implementation.

4.35 pm
Peter Minary [Oxford]

Title: Modelling by Hierarchical Natural Moves

Abstract: The computational modelling of biological structures is playing an instrumental role in expanding our understanding of biology at the molecular level. However, the full potential of biomolecular simulations relies on being able to predict biologically relevant collective molecular motion, which would spontaneously emerge only at time scales not always accessible by conventional molecular simulations. To address these limitations, here we introduce an algorithmic approach that supports modelling with customizable (hierarchical) degrees of freedom and opens up new application areas for modelling large macromolecular complexes.

5.00 pm
Jon Doye [Oxford]

Title: The oxDNA model and dynamics simulations of coarse-grained models

Abstract: OxDNA is a coarse-grained model of DNA developed in Oxford to study both large DNA objects and long-time scale processes in DNA. It has been mainly applied to study the fundamental biophysical properties of DNA (e.g. response to mechanical stress, the dynamics of basic DNA processes, such as hybridization or displacement) and the structure and self-assembly of DNA nanostructures (e.g. DNA origami). The model consists of effective interactions between rigid nucleotides representing chain connectivity, stacking, base pairing, electrostatics and excluded volume. As there is no explicit solvent, the model is simulated using dynamical algorithms that reproduce the diffusive dynamics of DNA in solution. In order to study systems with tens of thousands of nucleotides, a GPU-version of the dynamics code has also been developed. The talk will illustrate the power of this coarse-grained approach, but also highlight some of the difficulties when interpreting the dynamics. (The code is available at )

5.30 pm
6.00 pm
Off to the Royal Oak/Dinner