Research On Protein Complex Identification And Visualization In Protein Interaction Networks

With the continuous development of biological science and technology, large amounts of protein-protein interactions are explored. Protein interaction networks formed by these interactions become data sources of researches on protein function and protein complex identification done by biologists. Protein complexes consist of proteins with similar functions and affinitive relationships, therefore, present as dense sub graphs in protein interaction networks and are able to be identified efficiently by computational approaches from interaction networks. Visualizing protein interaction networks in process of analysis can intuitively present the organization of the network, which makes easy comparison of cluster results, promotes researches on protein interaction networks.In this paper, we focus on protein complex identification from protein interaction networks using graph clustering algorithms, and present research and implement of a protein interaction network visualization system.In the task of protein complex identification, various computational approaches have been proposed to detect protein complexes from protein interaction networks. However, high false-positive rate of protein-protein interactions makes the identification challenging. It is necessary to evaluate the quality of interactions in protein interaction networks, filter the interactions with low reliability to improve the performance of protein complex identification algorithms. We propose a protein semantic similarity measure based on the ontology structure of Gene Ontology terms and Gene Ontology annotations to estimate the reliability of interactions in networks. Interaction pairs with low GO semantic similarity are removed from the network as unreliable interactions. Then, a cluster-expanding algorithm is applied to identify complexes with core-attachment structure on the filtered network. We have applied our method on three different yeast PPI networks. The effectiveness of our method is examined on two benchmark complex datasets. Experimental results show that our method outperforms other state-of-the-art approaches in most evaluation metrics.In the task of protein interaction network visualization, we introduce existing network visualization systems, design and implement a protein interaction network visualization system with JUNG, a network modeling and visualization library. We also integrate some most effective clustering algorithms into the system, implement a detection method based on protein function annotation, and make the original network and cluster results better navigated with 2-dimension network views. Experimental results show that our system meets the requirements of protein interaction network researchers.