Virus as a whole: meso- and macroscopic structure and dynamics at all atom resolution
CECAM-HQ-EPFL, Lausanne, Switzerland
State of the art computers can simulate liquid molecular systems several hundred million atoms in size (tens of nanometres across) using classical molecular dynamics (MD) methodology. Experimentally, modern x-ray crystallography can measure atomistic structure of complete parts of living cells with up to ~0.1A resolution (for example, human ribosome , photosystem complex , entire virus [3,4]). These two facts together imply that it is now possible for the first time to model the structure of an entire virus at atomistic resolution. Even though at the moment there are no viruses whose complete structures are known (with the nucleic acid inside), it is a feasible task if the latest experimental data from cryo electron microscopy is used. However, simulating the dynamics of such systems over biologically interesting times is still impractical (up to 1µs simulation is reported ).
It has been recognised that water surrounding biomolecules (sometimes termed "biological" water) plays the major role in the dynamics of the biomolecule, so much as it should be considered as an integral part of the biomolecular system. The simulation of atomistic (explicit) water takes up to 90% of computing resources. Such modelling of viral processes over biologically relevant times (microseconds to milliseconds) is likely to remain a challenge for atom-resolved MD. Currently the only possibility to increase the simulated times is to model water as continuum in regions where the atomistic nature of water molecules is not critical (for example, in the bulk far from the biomolecules), which implies larger space and time scales. A fundamental problem is how to link the coarser continuum and finer atomistic representations.
A promising direction is the development of hybrid molecular dynamics-hydrodynamics approaches [6,7,8]. Such approaches consistently combine the atomistic and continuum representations at different parts of the simulated system. This will allow modelling of the virus itself and water near it at the necessary atomistic level and, at the same time, continuum water further away. The approaches also allow the simultaneous investigations of processes at space and time scale with orders of magnitude difference. This provides unique opportunities for studying biologically and medically crucial phenomena such as, for example, meso- and macroscopic transport of viruses carried by hydrodynamic flows in and between cells combined with the molecular specificities of their interactions at the microscale with the cell membrane and organelles.
The two main fields of molecular research will be brought together: 1) high performance molecular dynamics modelling of large biomolecular systems and 2) hybrid multiscale molecular dynamics-hydrodynamics modelling. Meetings in the two fields separately tend to focus on one aspect of the problem only. High performance large scale molecular simulation meetings usually attract biologists and biochemists interested in the biochemical aspects of the results. The methodological elements of such meetings typically concentrate on hardware developments and numerical techniques, leaving the fundamental physics of the methods of lower importance. The hybrid multiscale methods discussions often concern general physical and mathematical aspects of new approaches. For natural reasons, the applications are limited to small systems such as pure water or solutions of small molecules. We propose to combine the two approaches in a single meeting involving world leading experts from both sides. Computational biochemists will be able to appreciate the benefits of multiscale methods. Hybrid methods experts will benefit from the in depth knowledge of biomolecular systems and feasibility of their modelling.
More specifically, the following problems will be considered:
Reconstructing the structure of large biomolecular systems at all-atom resolution from lower resolution experimental data, such as cryo-electron microscopy densities.
Obtaining meso- and macroscopic properties, such as mechanical elasticity, diffusion and viscosity properties, thermal conductivity, of large biomolecular objects (whole organelles, viruses) from all-atom simulations.
Changing the number of degrees of freedom when coarse graining atomistic models from all-atom to mesoscale for simple liquids, solutions of small molecules, and large biomolecular systems.
Developing Langevin-like equations of motion for coarse grained models of large biomolecular systems, calculating the friction coefficients for these equations.
Consistently combining fine scale all-atom representation of the system with coarse grained representation of the environment and other parts of the biomolecular object.
Developing realistic models of molecular fluctuations when going downscale from macroscopic continuous hydrodynamics towards atomistic level of description, using Landau-Lifshitz Fluctuating Hydrodynamics approach as the basis.
Developing effective implementations of atomistic, continuous, and hybrid frameworks in high performance computation hardware, designing and building specialised accelerators for these algorithms.
Sergey Karabasov ( Queen Mary University of London ) - Organiser
Anton Markesteijn ( Queen Mary University of London ) - Organiser & speaker
Dmitry Nerukh ( Aston University, Birmingham ) - Organiser & speaker