Systematic assessment of protein interaction data using graph topology approaches

Abstract

Current protein interaction detection via high-throughput methods is highly erroneous. At the same time, the false negative rate of the interaction networks has also been estimated to be high. The purpose of this study was to investigate protein interaction networks from the topological aspect, and to develop a series of computational methods to automatically purify these networks, i.e., to identify true protein interactions from the existing PPI networks and discover unknown protein interactions by their topological nature. This thesis introduced three different approaches. First, it presented a measure called IRAP, and further IRAP*, to assess the reliability of PPI based on the alternative paths in the network. Second, the thesis presented a new model to identify true protein interactions with meso-scale network motifs. Third, the thesis exploited the biological information that are associated with network motif vertice to capture not only the topological shapes, but also the biological contexts.