Adaptive Multi-Resolution Massively-Multicore Hybrid Dynamics

We propose to develop highly scalable software that will exploit next generation, heterogeneous, massively parallel processors (such as those found in widely available graphics processors - GPUs) to deliver orders-of-magnitude performance increases for conformational sampling in molecular simulations. The software will be generally applicable to simulations of any condensed phase molecular system. The initial application area will be to accelerate the sampling of protein conformational change within the types of simulation used for rational drug design in the pharmaceutical industry.Future applications of rational drug discovery will depend critically on the ability to model protein conformational change and protein flexibility. Previous successful applications of computational methods in rational drug design targeted proteins that had small, well-defined binding pockets, in proteins that were either relatively rigid, or changed little upon drug binding. Increasingly, medicinally interesting protein targets have large, open and flexible binding sites. To understand binding, computational models have to be able to predict how these sites will change shape upon drug binding. Coupled to this, a new generation of drugs are being developed that target the interactions between protein surfaces, or that require modelling of protein-protein association. In these cases, the binding site is extremely dynamic, as it is formed between two (or more) proteins that have come together. Existing molecular modelling algorithms and software are incapable of stepping up to the challenge of modelling highly flexible proteins. New software and new algorithms are needed urgently to ensure that computational science continues to play an important role in the pharmaceutical industry.We have designed a new multi-resolution algorithm that will allow for the simulation of molecular dynamics to be broken into two parts; a near-field, atomistic part, and a far-field, coarse grain part. The near-field part is used to model the interactions between neighbouring molecules, using traditional atomistic forcefields, and uses a standard Monte Carlo (MC) algorithm to model the dynamics of individual atoms. The far-field part models the remaining molecular interactions using a coarse-grain (beaded) forcefield, and uses rigid-body dynamics to model global dynamics (e.g large-scale protein conformational change). This multi-resolution split of both the dynamics, and the modelling of the molecular interactions, makes the algorithm ideally suited to heterogeneous computing platforms such as supercomputers equipped with numerical accelerators (e.g. graphics processors). In addition, the software will also be energy-aware, as the energy cost of performing each part of the simulation will be factored into the decision as to which resource it is allocated. For example, if the results of the simulation were not needed immediately, then the simulation could be diverted from the accelerator, and instead run using low-power processors (e.g. clusters of Intel Atoms, like those found in netbooks). This would give the simulator the choice of minimising the total simulation runtime or the total CO2 cost. While developed for the clusters of today, the software will readily scale to the peta- and exascale supercomputers of tomorrow, where concepts such as software adaptability, energy management and fault-tolerance will be key to achieving efficient scaling and efficient supercomputer utilisation. We hope that one of the lasting impacts of this project will be a promotion of greater understanding of energy-aware algorithms and CO2/energy-aware scheduling in the international HPC community. Our intention is to tackle head-on the issues facing the international HPC community in increasing yet variable energy cost and availability, and the need to significantly improve the energy efficiency, and reduce the environmental cost of HPC.

The software developed during this research will be unique. It will be the first HPC multi-resolution hybrid dynamics program, and it will enable simulations of conformational change that had hereto been accessible only to a single specialised molecular dynamics program (Desmond) running on a unique, custom-designed special-purpose computer platform (Anton). The immediate impact of this software will therefore be to give scientists around the World the ability to run the equivalent of long-timescale molecular dynamics trajectories using widely-available graphics processors (GPUs), and/or using local or national HPC resources. The software building blocks developed during this project will be released as a separate library, enabling users in industry and academia to develop other novel hybrid dynamics programs. In addition, to broaden the impact to the wider community, we plan in stage 2 to port these building blocks into widely available molecular modelling frameworks (e.g. OpenMM) and will work with developers of existing dynamics software to see how these building blocks could be used to implement accelerated multi-resolution hybrid dynamics algorithms in existing codes. This will ensure that the impact and benefits of this software will be available to the widest possible international community of industrial and academic molecular modellers. The immediate beneficiaries of this software will be industrial and academic scientists interested in accounting accurately for protein conformational change within thermodynamic simulations, such as those used to predict protein-drug or protein-protein binding. This will be of great benefit to industrial scientists working in the pharmaceutical industry studying flexible protein-drug systems. The software and algorithms developed feed nicely into the International Exascale Software Project (IESP) of which SMS is an active member. IESP's mission is to stimulate the creation of new algorithms and methods that will scale to the first exascale systems due to go online in 2018. In addition, the energy-aware, fault tolerant queuing system developed as part of this project is highly novel, and would be of great use to international developers of HPC software. We will release this system as an open source library, with sufficient documentation to allow it to be used by other developers for a wide variety of HPC applications. The ability to create an energy/CO2-aware cloud computing platform provides a unique point of difference in an emerging and rapidly growing market. We would therefore seek to partner with cloud computing providers to allow commercialisation of the energy/CO2-aware cloud computing components, e.g. via licensing, or by forming a spin- off company that would handle support, provide access to cloud computing resources, and provide the enabled software (such as the hybrid dynamics software) via a software-as-a-service (SaaS) web-based portal. This project has the potential to put orders of magnitude more performance, performance per dollar and performance per watt into the hands of molecular modellers in industry and academia by significantly increasing the rate of adoption of GPU-based accelerators. The commercial and scientific benefits of this step-change in capability are potentially enormous for industry and academia. We hope that one of the lasting impacts of this project will be a promotion of greater understanding of energy-aware algorithms and CO2/energy-aware scheduling in the international HPC community. Our intention is to tackle head-on the issues facing the international HPC community in increasing yet variable energy cost and availability, and in tax regimes designed to encourage reduction in CO2 emissions, The long-term impacts will thus be on international HPC policy, and also in significantly improving the energy efficiency and environmental cost of HPC.