The Development Of Multi-level Interaction Network Expansion Technology For Traditional Chinese Medicine

The network pharmacology provides a new approach to study the active ingredients of traditional Chinese medicine and its mechanism. However, the number of compounds of traditional Chinese medicine and their targets has been reported is limited, which is unable to build complete multi-level interaction networks to completely understand and evaluate the efficacy of traditional Chinese medicine and its mechanisms.In this study, we take the treatment of cardiovascular disease by Fufang Danshen as model, to try to develop a known biomedical data based multi-level interaction network expansion technology for traditional Chinese medicine in order to improve the accuracy and effectiveness of network pharmacology in traditional Chinese medicine.This thesis focus on the data collection and network construction,expansion of the multi-level interaction network and the network analysis and comparison of prior and extended network, was divided into following four chapters:Chapter I IntroductionThis chapter first introduced the basic features of drug action networks, and then introduced the network analysis algorithms, in this part, we mainly introduced the node ordering and decomposition of network module based network analysis algorithms. The latter is based on the influence of nodes on the network structure and function, which can acquire more comprehensive and accurate information compared to the former. The third part of this chapter mainly focused on the present application of network analysis algorithms in traditional Chinese medicine. Due to the researches mainly build network of traditional Chinese medicine based on the limited known data, it is difficult to completely reveal the active components and its synergistic effect for a specific disease. Therefore, it is necessary to expand the drug action network by prediction. Thus, we introduced the prediction of active ingredients and targets in the forth part, including ligand structure based drug targets prediction,protein structure based drug targets prediction, and data mining based drug targetprediction. Finally, we introduced the active ingredients of Fufang Danshen and its mechanism of action.Chapter II The data collection for Fufang Danshen in treatingcardiovascular disease and network constructionIn this chapter, the data was collected for the construction of Fufang Danshen multi-level interaction network. Firstly, we collected and collated the compounds of Fufang Danshen, cardiovascular disease, protein(target) and their interaction data from the PharmGKB, OMIM, UniProt and other databases and literatures. Then, we unified their names, removed errors and duplicated data, eventually we got 201 compounds of Fufang Danshen, 34 cardiovascular related diseases, 494 proteins, 4389 pairs of Fufang Danshen components-proteins(targets) interactions, 78 pairs of Fufang Danshen components-cardiovascular interactions, 70 pairs of cardiovascular disease-protein interactions, and 48 pairs of proteins(targets)-protein(targets)interactions. Finally, we constructed a multi-level interaction network of Fufang Danshen based on the collected compounds, cardiovascular disease and proteins(targets).Chapter III The expansion of multi-level interaction network of FufangDanshenIn this chapter, we developed a typical correlation analysis prediction model based on chemical and protein molecular descriptors through combining the chemical(protein) space and typical correlation analysis model. The model can predict the interaction between multiple compounds and multiple targets, and the results of evaluation showed that the consistency of the model reached to 85.2%. Moreover, 463 pairs of compounds-protein(targets) interactions, 26 pairs of compounds-diseases interactions, 3830 pairs of proteins(targets)-diseases interactions were predicted by using canonical correlation analysis prediction model, and simple prediction which is based on the principle that the two nodes are related if they act on the same node, and the multi-level interaction network of Fufang Danshen was expanded.Chapter IV Network analysis and comparison of the multi level interaction network of Fufang Danshen in treating cardiovascular disease beforeand after the expansionIn order to evaluate the completeness and effectiveness of the expanded multi-level interaction network, network analysis and comparison of the prior and expanded multi-level interaction network of Fufang Danshen in treating cardiovascular disease were carried out through comparing the parameters of the network, comparing the disease centered network modules extracted from the prior and expanded network respectively, and comparing the functional modules obtained from the prior and expanded network by network analysis. The results showed that the prediction technique developed in this study can improve the completeness and accuracy of the multi-level interaction network of Fufang Danshen in treating cardiovascular disease.In conclusion, the existing biomedical data based prediction technique for multi-level interaction network expansion developed in this study can provide a basis for the evaluation and understanding of the active components and their synergistic action of traditional Chinese medicine in a more comprehensive and accurate manner.