Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.492255
Title: Frequency rule mining for effective protein-protein interaction inference from gene expression and protein structures
Author: Dafas, Panagiotis A.
ISNI:       0000 0001 3402 1510
Awarding Body: City University, London
Current Institution: City, University of London
Date of Award: 2008
Availability of Full Text:
Access through EThOS:
Abstract:
The experimental measurement of gene expression levels has produced preliminary results on the regulation, pathways and networks of genes in cells. Furthermore, the number of available three-dimensional folded structures of proteins increases on a daily basis. The ultimate aim of both genomics and proteomics froni the perspective of bioinformatics, is to map out all the circuits of energy and information processing in life by terms of molecular interactions in a system~tic way with minimal human intervention. In this thesis we propose a new rule mining framework for ill silico inference of protein-protein interactions and an effective class of techniques that identify domain-domain interactions from multiple domain protein structures. The above two approaches allow molecular biologists to use both gene expression data and three dimensional folded protein structures to efficiently predict or validate potential protein-protein interactions. A novel temporal rule mining technique is used to infer rules that relate local expression patterns across a set of genes. The set of these rules is associated to a set of potential interacting pairs of proteins. . I Probable protein-protein interactions can be validated against a network of protein domain interactions that are computed by considering interacting domains in known multiple domain protein structures. We introduce efficient algorithms that can effectively identify protein domain-domain interactions in multiple protein domain structures which are solved and publicly available. To analyse the vast amount of interactions between all the protein domains classified to superf~milies we propose a new graphtheoretic measure which is able to rank protein superfamilies by incorporating information about the topology of the whole network. To summarise, the work in this thesis consists of a new temporal rule mining framework applied .' to gene expression data analysis and furthermore, a novel class of algorithms that effectively identify ' domain-domain interactions from known protein structures.
Supervisor: Not available Sponsor: Not available
Qualification Name: City University, London, 2008 Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.492255  DOI: Not available
Share: