Research On Drug Target Identification And Repositioning Based On Protein Interaction Network

The discovery of drug targets provides the foundation of drug development. Although the complexity of causes resulting in most diseases may make these disease-related genes or proteins be potential drug targets, the number of novel drugs haven't increased equally with rapid progress in'-omics'technology, such as genomics and proteomics, because of the robustness of biological network, the specificity of potential drug targets and other underlying factors. With increasing high-throughput data, the development of systems biology makes it possible to integrate information of drugs, protein interactions, disease phenotype and other related data. This has promoted the emerging of network pharmacology which provides a network perspective on understanding of diseases and drug target discovering. It's still an open and essential question on how to discover novel drug targets and develop new strategies of drug repositioning based on molecular network.Here, we have aimed at the need of anti-radiation drugs, anti-tumor drugs and host-targeted antiviral drugs, and studied key issues of computational analysis of drug-target discovery and drug repositioning based on protein-protein interaction network. We first proposed a new way of identifying active subnetwork by integrating gene differential expression and correlation expression, then this method was applied to analyze disease networks on radiation damage and EV71 infection, the potential drug targets were experimentally validated. Secondly, we implemented a tool on dynamical analysis of network perturbation. It's then applied to study the combination perturbation and propagation of perturbation in quantitative medulloblastoma network. Finally, we discussed the potential antiviral ability of current drugs through data integration of drug-targets, essential host factors for virus infection and virus-host protein interactions.Causes of complex diseases rarely resulted from a single gene mutation, and often related to many systematic changes in the underlying cellular networks. Network pharmacology presents a complex network view and helps to relate disease and drug in broad ranges. In the first chapter, we introduced the background of drug target discovery, especially antiviral drug targets, and drug-repositioning. The critical issues in these areas are discussed. We then posed questions and developed our technical route. In the end, the main research contents and organization of this paper are presented. The footstone of network pharmacology is to construct related molecular network of specific diseases. It's has been proved that disease networks could be generated through integration of gene expressions and protein-protein interactions. In second part, we proposed an integrated index to search for active subnetwork. Step iterative algorithm was established to identify significantly differential and strong correlated expressed genes simultaneously. This method also could recognize those key genes with low expression. The validity of the method on disease networks was further confirmed When applied to analyze gene expressions of radiation damage and viral infection. The results were consistent with former reported studies. With further experimental validation, several proteins were identified which present potential effects of anti-radiation and anti-virus infection.Protein interaction networks are essentially dynamical. The dynamical characteristics of proteins in the network are often more important than static topology ones in drug target analysis. In the third part, we implemented the dynamic model of network perturbation, and developed a tool to integrate data of quantitative proteomics and protein interaction network. The tool was then applied to analyze combination perturbation and applied to medulloblastoma network. The results indicated that propagation of protein concentration changes exponentially decreased. We also found that intensity of protein perturbation is not necessarily correlated with its concentration and topology. Further analysis by considering drug targets revealed that high perturbation proteins have shorter distance with drug targets in the network with respect to hubs. These results showed the perturbation model is close to the real situation of drug action, which will have an implication on drug target discovery and drug repositioning.Viral diseases cause serious harm to human health. Traditional antiviral drug development mainly focused on viral proteins, which lead to drug failure and resistance due to high mutation in viruses. The long cycle of a successful drug development can't keep with sudden outbreak of viral diseases and satisfy needs of disease control and prevention. In the fourth part, we first presented a systematic analysis of relation between current drugs and viruses. Based on the common human targets, a drug-virus relation network was constructed consisting of 89 types of viruses and 1392 different drugs. Through calculating the shortest distance between drug targets and virus targets, we found that most virus targets are covered by drug targets or have direct interactions with drug targets. Finally, target topology characteristics of drugs, viruses and diseases are computed, and potential broad-spectrum antiviral targets and drugs are listed. This work has provided first analysis on potential antiviral strategy of current drugs and a new way to develop host-targeted antiviral drugs.In the final part, a brief summary on the work was introduced, and further work is prospected based on current research trend.