ERCIM News No.45 - April 2001 [contents]
UK e-Science Grid Programme
by David Boyd, Paul Jeffreys, Julia Goodfellow, Lois Steenman-Clark and Peter Allan
In October 2000, the UK Governments Office of Science and Technology (OST) announced a £98 Million programme over 3 years to develop and deploy Grid technology to meet the data and computing challenges arising in several areas of the physical and life sciences.
The rationale behind this decision is that, today, science is posing some of the most extreme computing challenges. These challenges represent the cutting edge in developing the next generation of distributed IT systems. Just as happened with the Web, however, the solutions to these challenges will quickly find application in the commercial marketplace. By supporting the Grid developments required by science now, the belief is that we will get to commercially useful systems quicker.
This new funding is being channelled through the current structure of the UK Research Councils which support basic and applied research in engineering and physical sciences (EPSRC), particle physics and astronomy (PPARC), the natural environment (NERC), biotechnology and biological sciences (BBSRC), medical science (MRC) and economic and social science (ESRC). CLRC, the seventh Research Council, provides major scientific research facilities in support of the work of the other Research Councils and also supports UK scientists in their use of large scale facilities elsewhere in the world. Together, all these facilities will generate increasingly large quantities of increasingly complex data over the next decade. Grid technology offers a potential solution to the challenge of extracting useful science from this complexity.
Some of the developments which are proposed as part of the UK e-Science Grid programme over the next 3 years are summarised below.
Molecular simulations are an essential tool in understanding macro-molecular function at an atomic level. Researchers at Birkbeck College in London, supported by BBSRC, have been using molecular simulation techniques, such as molecular dynamics, to study the stability and unfolding of specific proteins where the onset of partial unfolding is correlated with disease. One example of this is the protein human lysozyme (see figure) for which two mutations have been found clinically associated with amyloidosis (formation of the protein into fibres resistant to temperature and enzymes). These calculations are computer intensive and have been traditionally carried out on supercomputers and more recently on in-house departmental multiprocessor workstations. Recent work using a Beowulf cluster at the Rutherford Appleton Laboratory (RAL) of CLRC has demonstrated this architecture to be a very cost effective solution to the considerable computing requirements of these techniques. A project funded by BBSRC and the Wellcome Trust is establishing a prototype Biosciences Grid linking five molecular simulation groups at Oxford, Southampton, Birmingham and York Universities and Birkbeck College with RAL who will provide technical support and computational and data resources.
Structure of a human lysozyme mutation.
Earth System Science
One of the possible proposals for a major new NERC initiative has the goal of predicting environmental changes on timescales of up to 100 years on the basis of:
- a sound scientific knowledge of the processes within, and the interactions among, the components of the coupled earth system
- an understanding of the uncertainties of our knowledge of key processes
- the inherent limitations of the predictability of the system.
The intention is to develop a modular modelling environment in which a hierarchy of models can be supported in a common framework. The project would involve collaboration between a broad spectrum of UK researchers from many areas of climate research and earth observation. The primary aim of this prototype climate research Grid is to provide the Earth System Science community with:
- a Data Grid providing high performance, integrated access to distributed massive data resources
- a Simulation Grid supporting large scale and distributed modelling capability spanning process studies to fully coupled climate change experiments
- a Model Grid providing a modular model code framework supporting multi-model experiments.
Astronomyop observatory to collect new data. These federated resources will contain data, with appropriate metadata to support efficient search and retrieval, from a wide range of high resolution, all-sky, multi-waveband surveys. Among the data sources feeding into the AstroGrid will be the UK Infra-Red Telescope Wide Field Camera (UKIRT-WFCAM) on Mauna Kea in Hawaii, the Sloan Digital Sky Survey on Apache Point in New Mexico, the XMM-Newton X-ray satellite and the Cluster mission of four satellites measuring the effect of emissions from the Sun on the earth's magnetosphere (see front cover picture). As an example of the productivity of these facilities, UKIRT-WFCAM can generate a substantial fraction of a terabyte of data every clear night.
The UK astronomical community is planning to set up a Grid based project called AstroGrid. This will involve the astronomy centres with major data archives at RAL, the Mullard Space Science Laboratory and Edinburgh, Cambridge and Leicester Universities, and will bring in computer science expertise from Queen's University Belfast. Naturally there are important astronomical resources elsewhere and all of these will eventually be accessible to the whole UK astronomical community via the Grid. The concept of AstroGrid is to create a Virtual Observatory. This will enable researchers to obtain new knowledge about a particular class of astronomical objects by extracting and analysing information from many existing data resources distributed throughout the world, rather than by applying for scarce telescope time at some remote mountain
David Boyd - CLRC
Tel: +44 1235 446167