In this post-genomic era, biological networks are crucial resources for the next generation of bioinformatics research. The systematic analysis of biological networks, such as protein-protein interaction networks and gene regulatory networks, has provided a new paradigm for characterizing functional behaviors of genes and gene products. Recent advances of high-throughput experimental techniques have generated protein-protein interaction (PPI) data on the scale of the entire genome, collectively referred to as the interactome. The availability of interactome data has catalyzed the development of computational approaches to elucidate protein functions on a system level. However, these approaches have been challenging because of a significant amount of erroneous data and complex connectivity of the networks. In this talk, I present novel computational approaches for functional pattern mining from PPI networks. Unknown protein functions or functional modules can be predicted based on their linkage patterns hidden in the PPI networks. My approaches also employ the integration of topological features of PPI networks with semantic analytics of proteins in order to improve accuracy. Hierarchical organizations of proteins can be detected using an information propagation model. Potential signaling pathways can be identified by mining frequent paths in the PPI networks. High efficiency and scalability of these approaches demonstrate that they are well-applicable to high-level organisms such as human. This study will provide a useful framework for various biomedical applications.