BioSysBio:abstracts/2007/Alastair Spence


 * Add or delete the sections that you require.

Assessing cross-talk in biological networks - a sensitivity analysis of spectral clustering
Author(s): Alastair Spence, Zhivko Stoyanov, J Keith Vass Affiliations: University of Bath The Beatson Institute for Cancer Research Contact:email: as@maths.bath.ac.uk Keywords: 'add-keyword_1' 'add_keyword_2' 'add_keyword_3' 'mRNA-Splicing'

Background/Introduction
Large networks have been obtained on the flow of information from DNA to proteins. Two techniques are presently providing large datasets – microarrays (messenger RNA levels) and protein binding studies. Here we analyse microarray data where clustering is an important analytical tool - often achieved using spectral analysis of the Laplacian or adjacency matrix of the graph of the network. These clustering techniques are heuristic and it is natural to ask how reliable they are. In this work we provide a sensitivity analysis, based on the use of numerical linear algebra and probability analysis to obtain a measure of the robustness of spectral clustering.

In biological networks components have more than one function, known as "cross-talk”; as a paradigm for this we analyse apparently linked alternative splicing events in the Transforming Growth Factor Beta (TGFB) pathway. Here the role of common elements is altered by the presence or absence of regions of messenger RNAs (mRNA). A number of mRNAs, with linked biological function, are studied with an analysis of the likelihood that parts of the mRNA are present or absent in a cluster.

Results
Add your text here

Images/Tables
Add your images or tables here

Materials/Methods
Add your text here

Conclusion
Add your text here