A Specialised Computational Resource for Biomolecular Simulation

Molecular dynamics is a computer simulation method that allows one to visualise the ways in which molecules change shape, whether that is just as a result of thermal energy or as a result of interactions - maybe with other proteins or with small molecules, for example drugs. These changes in shape are critical for the ways in which biomolecules 'work', but at the moment no other experimenatl method allows us to styudy this clearly at the atomic level. Understanding molecular dynamics is vital if we are to explain how cell signalling works (and so for drug design), how cells, and ultimately whole organisms, move, and how to build artificial 'molecular machines' for nanotechnological applications.

Research in the Division of Medicinal Chemistry and Structural Biology in the School of Pharmacy at Nottingham involves the close integration of information about the structure, dynamics and recognition properties of biomolecules (particularly proteins and nucleic acids) into a wide range of projects related to drug discovery and biomolecular engineering. One of our core methodologies is molecular dynamics simulations, which can provide insights into the structure, dynamics, and recognition properties of biomolecules that are not available by any other experimental technique. These simulations are extrememly compute intensive, jobs may run for weeks or even months at a time, and for maximum throughput require parallel processing architectures with very good interconnects to optimise performance. This application is to replace a key component of our resource, a 36-processor Beowulf cluster obtained through a JREI bid in 1999-2000 with new hardware. This will be used to facilitate a range of current research projects, ranging from the design and development of novel antitumour agents, to fundamental studies of protein signaling and solvation, to simulations of molecular deformation in single-molecule experiments. With our commercial partners we will also be initiating projects to investigate how emerging networking products can optimise the performance of MD codes on parallel architectures, and how tools for high performance remote visualisation can be adapted to the needs of the wider biosimulation community to share data with each other, and with researchers in other disciplines.