Topological Analysis of Biological Networks

Declaration I hereby declare that this thesis is entirely my own work except where otherwise indicated. I have used only the resources given in the list of references. Abstract An astonishing diversity of biological concepts and processes can be represented by the relatively simple formalism of networks, which consist of nodes connected by edges. Examples range from neuron communication to metabolism and gene suppression mechanisms. The pursuit of describing many interactions and dependencies between the constituents of a living organism has produced large networks of interacting proteins, coexpressed genes, metabolite transformations and other. With advances in large-scale experimental technologies, such as microarray experiments and yeast-two-hybrid screens, data are constantly increasing both in size and complexity. Additional techniques, such as literature curation and prediction methods diversify the biological data in the available repositories. The development of a unified framework for the analysis and integration of these data has appeared as a central challenge in bioinformatics research. This work presents two applications for the manipulation and investigation of biological networks. The Cytoscape plugin named NetworkAnalyzer facilitates topological analysis of large networks through computation and visualization of a variety of network parameters. Simultaneous inspection of analysis results of two or more networks enables network topology comparison. The stand-alone tool NetworkLoader compiles interaction networks from different molecular interaction datasets, providing flexible criteria for data filtering and integration of identifiers. The successful application of NetworkLoader and NetworkAnalyzer in handling protein-protein interaction data, as well as community demand for the functionality provided by NetworkAnalyzer encourages future support and development of these applications. In addition, we introduce an adaptable approach for quantifying the topological role of a subset of nodes in a network. The approach consists of estimating p-values for a collection of topological properties., It application to groups of polyanion-binding proteins within protein-protein interaction networks in yeast and human revealed intriguing topological features of some of the analyzed groups.