Disease specific gene expression profiles in the context of molecular networks

Abstract

Molecular Biology research has seen significant advancements in experimental technologies, frequently denoted as the 'Omics' revolution.<br />However, successful strategies for the integrative analysis of these growing data sources have remained an obstacle. Major challenges include integration efforts of the various Omics tracks, followed by their interpretation in the context of complex cellular processes as e.g.<br />comparing healthy and diseased phenotype. Such an integrated view on Omics data has centrally nurtured the field of Systems Biology.<br />Individual Omics data sets are, by their nature, of large size, highly heterogeneous, leaky, and exhibit some degree of error, all embedded in high dimensionality. These boundaries set the challenge for computer science towards identifying solutions in the realm of computational Systems Biology. This thesis is aimed to contribute to the methodological as well as application aspects of computational Systems Biology to understand disease conditions with a particular focus on cancer. Central ingredients for this approach come from disease associated gene expression profiles, followed by their interpretation in the context of protein- and genetic interaction networks.<br />The methodological aspects of my work discuss an Omics data integration pipeline providing a protein-centered interaction network embedding the entire, presently annotated human proteome. This reference interaction network can now be used for analyzing Omics profiles characterizing specific cellular conditions. I present analysis examples on gene expression profiles characterizing B-cell lymphoma, breast cancer, and kidney diseases.<br />Our approach of analyzing gene expression profiles on the level of interaction graphs successfully identified lymphoma associated consensus subnetworks, yielded synthetic lethal hubs linked to downregulated features in breast cancer, and identified steroid targets for overcoming acute inflammation in kidney transplant organs.<br />Integration efforts as presented in this thesis provide the basis for any Systems Biology analysis of complex cellular states and advancements of experimental technologies - as e.g. presently seen with next generation sequencing - will at high rate add further qualitative and quantitative data. From this the importance of computational analysis procedures will further rise for effectively analyzing and interpreting Omics data in their phenotype context.