edikt2010 Symposium - Using computing in your research
Wednesday April 28th, 2010
e-Science Institute, 15 South College Street, Edinburgh
Poster Abstracts
Dr George Beckett
EPCC, University of Edinburgh
DiGS, distributed-data management
A significant challenge for many research groups is how to manage the huge volume of raw data that is generated by their experiments, simulations and imaging. Applications from biological sciences, particle physics, and astronomy - to name a few - commonly generate terabytes of data, which is both unwieldy to manage and expensive to process. The problem is compounded by the distributed nature of today’s scientific communities. Researchers from different institutions - and different countries - regularly collaborate on particular projects, and thus need to share data. To meet these challenges, a team of software engineers from EPCC has developed DiGS; a distributed, data management system that leverages significant technological innovations from grid computing. DiGS supports the management, sharing, publication and preservation of collections of data. It is specifically designed to handle large volumes of scientific data on dispersed, storage facilities, and furnishes semantic data management through the binding of application-specific metadata to datafiles.
Mr Keith Evans
Department of Physics and Astronomy, University of Glasgow
Computer Modelling of Amorphous Structures for Advanced Gravitational Wave Detectors
We are working on modelling the structure of amorphous materials for the purposes of understanding their structural properties. To facilitate this process we use a combination of Reverse Monte Carlo and DFT methods to create a model with a reduced Density function (RDF) which matches experimentally produced RDFs, this results in a theoretical structure which closely emulates the experimental structure.
This poster will explain this method in greater detail and show some of the results obtained so far.
Dr David Rodriguez Gonzalez
SINAPSE Collaboration, NeSC/SBIRC, University of Edinburgh
PrivacyGuard: A DICOM De-Identification Toolkit
PrivacyGuard is an open-source DICOM de-identification toolkit implemented in Java. It is an extensible toolkit that allows users to easily write their PrivacyPolicies using a GUI and also to implement their own anonymisation methods in Java. Additional operations can be easily added including: data transfer (via SFTP or DICOM), cataloguing, etc. It also supports the DICOM standard attribute-level confidentiality mechanism.
Matthew Hartfield
Institute of Evolutionary Biology, University of Edinburgh
The role of advantageous mutations in enhancing the evolution of a recombination modifier
Recombination can evolve by breaking down interference among multiple loci. This leads to selection on a recombination modifier in a population subject to recurrent deleterious mutation. We show that if advantageous mutations are present as well, this significantly increases modifier selection strength. However, the strength of selection on a modifier is less than the summed strengths had there been deleterious mutations only and advantageous mutations only. Recombination always increases variance in fitness, which can improve population response to selection (see Weismann’s theory for evolution of sex).
Dr Carole Morrison
School of Chemistry, University of Edinburgh
What is the mechanism of proton transport through an alpha-helical channel?
This poster reports on quantum mechanical modelling work to ascertain how excess protons pass through a bundle of hydrated alpha helices. This process is of fundamental importance in chemical biology, giving rise to many basic functions including pH regulation, cell signalling and bioenergetics.
Our results indicate that the mechanism, whilst sharing some similarities with the accepted mechanism for proton transport in bulk water, also has important differences. A new mechanism for proton transport through a channel environment is therefore proposed.
Dr Stuart Purdie
University of Glasgow
gqsub - Making using the Grid really boring
gqsub implements the familiar qsub interface, making it trivial for users familiar with cluster computing to use the Grid. It aims to allow the same mechanisms and syntax for job control, so that the same job can be run on a local cluster, or the Grid, without modification. This facilitates using the Grid as an overflow mechanism for an existing cluster, as well as reducing the time to learn to use the Grid.
Mr Tony Weir
Information Services, University of Edinburgh
Edinburgh Compute and Data Facility
The Edinburgh Compute and Data facilities (ECDF) provides research computing services to all researchers at the University of Edinburgh and associated institutions, and to their collaborators.
Primarily ECDF delivers a central Linux compute cluster of 1456 CPU cores, with a high bandwidth, low latency Infiniband network connecting 256 CPU cores.
Storage and data backup services are also available.