1. Association of variations in I kappa B-epsilon with Graves' disease using classical and my Grid methodologies Peter Li School of Computing Science University of Newcastle upon Tyne

2. In silico experiments in bioinformatics Bioinformatics analyses - in silico experiments - workflows EMBL BLAST Clustal-W Genscan Resources/Services Example workflow: Investigate the evolutionary relationships between proteins Clustal-W Protein sequences Multiple sequence alignment Query

3. Issues in bioinformatics Large amounts of distributed data in different databases Highly heterogeneous Lack of standards Many applications Different algorithms, implementations Lack of programmatic access Difficult to perform bioinformatics analyses

4. e-Science and my Grid e-Science for in silico experiments –Potential solution to problems in bioinformatics –Using the Grid as a framework –Bioinformatics resources deployed as Web Services my Grid –Development of middleware to support the performance of in silico experiments in biology –Investigated the use of my Grid workflow technology in the genetic analysis of Graves’ disease

5. Graves’ disease Autoimmune thyroid disease Lymphocytes attack thyroid gland cells causing hyperthyroidism An inherited disorder Complex genetic basis Symptoms: –Increased pulse rate, sweating, heat intolerance –goitre, exophthalmos

6. In silico experiments in Graves’ disease Microarray data analysis Gene annotation pipeline Design of genotype assays for SNP variations

7. Classical approach to the bioinformatics of Graves’ disease Data Analysis - Microarray Import microarray data to Affymetrix data Mining Tool, Run Analyses and select Experiment Design to test Hypotheses Find restriction sites and design primers by eye for genotyping experiments Study Annotations for many different Genes Select Gene and Visually examine SNPS lying within

8. Taverna workflow system Used to compose and enact in silico experiments in my Grid Freefluo enactor Scufl language Workbench GUI –Service browser –Model explorer for workflow composition –Graphical view of workflow Free and open source http://taverna.sf.net

9. Modelling in silico experiments as workflows Semantic, syntactic and format typing of data in workflow Data has to be filtered, transformed, parsed for consumption by services Annotation Pipeline GOMEDLINEKEGGSwissProtInterProPDBBlastHGBASE Query

10. Results: Differential expression and variations of the I kappa B-epsilon gene Mean NFKBIE expression levels - Controls: 1.60 +/- 0.11 (SEM) GD: 2.22 +/- 0.20 (SEM) P=0.0047 (T-test) n=30 3’ UTR SNP – 3948 C/A - Mnl restriction site - χ 2 = 9.1, p = 0.0025, Odds Ratio = 1.4

11. Comparison between conventional bioinformatics and Taverna workflow approaches Advantages –Graphical composition of experiments –Automation and speed –Management of workflow information –Share and reuse workflows for other diseases Issues –Initial cost of learning/activation energy –Lack of Web Service interface to required resources

12. Acknowledgements Institute of Human Genetics –Simon Pearce and Claire Jennings School of Computing Science –Anil Wipat, Matthew Pocock and Keith Hayward European Bioinformatics Institute –Tom Oinn