Data integration methods for systems-level investigation of gene functional association networks

In this thesis, I discuss three studies that incorporate network-based approaches for the integration of heterogeneous biological data sources. I describe several novel statistical-based data integration methodologies, and demonstrate their utility in the discovery of new gene functional associations, as well as their ability to offer high-level insights into a variety of biological systems. The methods described herein operate on networks consisting of nodes that represent genes or gene products, with links between nodes indicating interactions identified using different high-throughput data sources. I present the application of these methods to the characterization of synthetic genetic interactions, the comparison of gene knock-down phenotypes, and the de novo discovery of archaeal transcription factor binding sites. In total, these three studies span multiple levels of network analysis, including building an analysis from the ground up starting with experimental data or genomic sequence, and the evaluation of similarity metrics for the comparison of biological data. The results of this work provide a first glimpse into the role of synthetic genetic interactions in a metazoan, a novel application of information-theory based similarity metrics to gene knock-down phenotype comparisons, and insights into global characteristics of archaeal transcription regulation. Furthermore, these studies as a whole demonstrate the utility of network-based data integration approaches for the study of biological problems.