Structural Property Analyses And Automated Regeneration Of Large-scale Signaling Networks

Researches in signaling networks contribute to a deep understanding of organism living activities. With the development of experiment methods in the signal transduction field, more and more mechanisms of signaling pathways have been discovered. Analysis and utilization of such signal transduction data become a challenge for biologists. Here, we tend to explore the signaling pathways by the bioinformatics approach.Firstly, we introduced the basic components, universal mechanisms and characteristics of signaling pathways. We summarized the current researches and latest progress in the area of bioinformatics analyses of signaling networks, including database resources, the structural properties analysis, automated pathways regeneration, the modeling and simulation of signaling networks.Sencondly, measuring each protein's importance in signaling networks helps to identify the crucial proteins in a cellular process, find the fragile portion of the biology system and further assist for disease therapy. Howerver, there are relatively few methods to evaluate the importance of proteins in signaling networks. Therefore, we developed a novel network feature, SigFlux, to evaluate the importance of proteins in signaling networks. Significant correlations were simultaneously observed between SigFlux and both the essentiality and evolutionary rate of genes. Further classification according to protein function demonstrates that high SigFlux, low connectivity proteins are enriched in receptors and transcriptional factors, indicating that SigFlux can describe the importance of proteins within the context of the entire network.Thirdly, signal flow direction is one of the most important features of the protein-protein interactions (PPIs) in signalling networks. However, almost all the outcomes of current high-throughout techniques for PPI mapping are usually non-directional. Based on the pairwise interaction domains, we defined a novel parameter Protein Interaction Directional Score (PIDS) and then used it to predict the direction of signal flow between proteins in proteome-wide signaling networks. We took the protein interactions with known directions in human, mouse, rat, fly and yeast as the golden standard positive set and the non-directional protein complexes as the golden standard negative set. Using 5-fold cross-validation, our approach obtained a satisfactory performance with the accuracy 89.79%, coverage 48.08% and error ratio 16.91%. Simulatiously, we presented two other approaches to predict the signal flow in pairwise protein interactions, which are the function based on GO function annotation and the support vector machine based on protein sequence. The accuracy and coverage are improved.Fourthly, considering the different characteristics of the above three methods, we proposed a bayesian network to integrate multiple data sources and gave likelihood ratio to evaluate the direction of signal flow between proteins. The integrated approach performed better than any prediction method based on individual source, with higher accuracy and coverage. Taking the proper threshold of likelihood ratio as 16, the bayesian method can achieve the accuracy 98.64% and coverage 67.83% in human protein interaction dataset. Simultaneously, to facilitate this strategy used by the community, we presented a web server to compute likelihood ratios of given protein interactions that allows users to infer signaling pathways from their interacting proteins.Finally, we applied the methods to the integrated human protein interactions and established a Directional Protein Interaction Network (DPIN). The DPIN is composed of 5,111 proteins and 10,051 interactions, and involves a large amount of novel signaling pathways. The DPIN was strongly supported by the known signaling pathways literature (with the 89.23% accuracy), and further analyses on the biological annotation, subcellular localization and network topology property. All these methods we proposed could be applied to predict signal flow direction in proteome-wide protein interactions and provide a global directional annotation of the protein interaction network. These methods are powerful not only in defining unknown direction of protein interactions, but also in providing comprehensive insight into the signaling networks.In summary, we elaborated on the signaling networks using bioinformatics methods in this paper, from the importance evaluation of each protein to automated regeneration of signalling pathways, providing a new understanding to the mechanism and principle of signaling networks. These methods can be applied to assist experiment design and drug discovery, and have a good prospect for development and application.