Molecular networks pose a challenge to biomedical research due to the complex interaction patterns and the comparably slow progress in identifying and characterizing the biological entities involved. Advances in high-throughput technologies for identifying and quantifying biological entities relevant for describing clinical phenotypes have narrowed the data gap over the last years, resulting in a large amount of heterogeneous biological data. Analysis of this data landscape in a molecular mechanistic context, supported by data management and knowledge discovery technologies, has in the meantime provided network based approaches for discovery of novel diagnostics and therapeutics for addressing unmet clinical needs. This work presents and discusses extensively the advances in molecular network analysis specifically in the realm of Systems Medicine over the last years and gives an overview on network principles, data sources and representation standards and analytical methodologies. Scientific contributions presented in this thesis address specifically (i) modeling of protein-protein interaction networks, (ii) tackling the false negative discovery rate of explorative omics experiments by proposing concepts of predicting relevant biological entities based on graph expansion algorithms and (iii) the practical analysis of omics data utilizing network-based methodologies. The proposed prediction concepts for biological entities take into account our current knowledge of the topology of molecular networks, and are based on vertex neighborhood as well as minimum spanning trees. As a result, a novel algorithm was developed for efficiently calculating a connecting spanning tree over a subset of selected vertices of a graph with low but not necessarily minimal overall edge weight. These prediction concepts as well as other network based methodologies were then applied to the analysis of omics data characterizing ovarian cancer, mesothelial cells stress response, and angiogenesis in brain metastasis, providing additional insight regarding their underlying molecular mechanisms. The complexity of network analysis remains challenging, nurturing research in life sciences as well as computer science. While the analysis of omics data with network methodologies sheds light on molecular mechanisms, the insights gained on natural network topologies and mechanisms further the design and analysis of networks in general. Analysis concepts and algorithms presented in this work serve as general approaches for omics data integration and analysis in the biomedical research area.