In recent decades, huge amounts of data have been accumulated in biological science and medical science. Under the guidance of reductionism, biological researchers focused on the study at the molecular level, yielding substantial findings on genes, proteins, RNA and metabolites. Meanwhile, clinical medical records also undergoing rapid growth. The immense accumulation of relavant data is both opportunity and chanllenge to the researchers. How to find valuable knowledge from these data is a crucial problem, and knowledge discovery is an efficient strategy. At the same time, researchers gradually become aware of the complexity of biological systems, and find that biological network is quite suitable for the knowledge discovery on biological systems. In this study, we are dedicated to research on data mining algorithms, which are critical in knowledge discovery, on the basis of biological network. The proposed algorithms were applied in the field of Traditional Chinese Medicine (TCM), in order to extract valuable knowledge which is beneficial to the modernization of TCM. The research contents of this study are summarized as follows: (1) As TCM usually contains numerous chemical compounds and exerts a complex mechanism of action, we propose a Pathway Pattern-based method for the prediction of active components of TCM. By utilizing bidirectional strong association rule mining algorithm, we firstly extracted the Pathway Pattern, which is made up of groups of pathways. Then the Pathway Pattern was used to prioritize chemical ingredients and gene targets through designing scoring functions. The novel method was applied in maxingshigan-yinqiaosan Formula, in which case 16 active components and 29 gene targets were identified. The prediction results were subjected to experimental and literature validation. By comparing to previous literature findings, we demonstrated the top ranked genes’ roles in the pathogenesis of H1N1 influenza. Further, molecular docking was utilized to validate the compounds’ effects through docking compounds into drug targets of oseltamivir. Finally, an active component-gene target interaction network was acquired to elucidate the pharmacology of maxingshigan-yinqiaosan formula. (2) Based on clinical medical records from traditional Chinese physicians, we propose a two-level model for the analysis of TCM syndromes. Firstly, a diagnosic model was generated. We selected core symptoms by using information gains, and generated a core s...
修改评论