Study On Circuit Regulatory Network Construction And Its Applications

At present, with the development of genomics and the corresponding improvements in technology, the expression data is not only about gene, also including the methylation data, miNRA data, protein data. The technical innovation and the increase of the amount of data allows us to explore the basic genomics task is no longer considered a single gene regulatory functions. Gene controlled by different regulators in different levels of regulation, and the organism is a complex organic, the presence of genes in the living body is not an isolated gene and not a simple linear relationship too. The interaction between genes and proteins, the interaction between genes and interaction between organisms and the external environment is very complex, they form a complex network system. To reserch of the living body and its behavior is difficult to understand for a single gene. For complex biological networks in transcriptional regulation and post-transcriptional regulation of the function and meaning of the stage acquisition, this study will use existing biological networks, bioinformatics algorithms and tools to build a transcription factors, miRNAs and genes makeup interactive regulatory network (Circuit). To know the overall circuit regulatory networks, can help us better understand the biological gene transcription, even more interesting translation regulatory mechanism; Secondly, from the changes in the structure of the circuit regulatory networks, it help us understand the biological phenomena such as cell differentiation and cancer generated understanding, to give research to provide the theoretical basis and support of these biological phenomena; once again, help medical experts found that the control and treatment of cancer from a pathological point of view. Therefore, this paper proposes a continuous-time point mode evaluation criteria, combine the method based on association rules mining circuit regulatory networks, consider a different circuit dynamic regulatory processes in the mouse lung development process. On the other hand, consider the difference between the normal samples and cancer samples, further analysis of the different features of the circuit regulatory networks. Based on the above analysis, we can see that the circuit is a very meaningful control unit, and has its unique characteristics between cancer and normal samples, combined with feature selection algorithm, to design a support vector machine classifier, do the cancer prognosis analysis.From the following aspects of the circuit construction of regulatory networks and the specific application.(1) For the evaluation of the expression continuous timing data. Expression data for consecutive less time points correlation measure of its intention to carry out the following studies:the concept of mode change posture according to the trend of expression changes between successive time points analyzed, the design of high expression and low principles for determining the expression construct a perfect tree structure exhaustive expression of trends. For the diversity of the gene expression data, considering the different evaluation criteria, select the appropriate evaluation criteria, describe the regulatory relations.(2) Design for less at the point-in-time pattern mining algorithm. Design association rule mining algorithm, co-regulated mode mining successive time points to determine the relationship between the mode, dig out a meaningful mode We intend to study. For the different number of point-in-time mode, to achieve the maximum associated mode, optimizing the largest association mining algorithms, including the setting of the threshold, the search rule-making, pruning the formulation of a strategy to develop a heuristic search algorithm optimized.(3) Based on the mode of mouse lung development data circuit dynamic regulatory network constructed instance. On the basis of the database, known transcriptional regulation and post-transcriptional regulation of the circuit network is constructed based on the relationship between mode, combined with timing of the development of the lungs of mice chip data using computational means, the study could influence factors of the development process of the mouse lung.(4) Case studies based on the mode of mouse lung development and the development of cancer the circuit dynamic control mechanism. Mining adverse gene expression trends in cancer samples with normal sample expression will focus on analyzing the factors of the abnormal expression of these genes, to explain its biological mechanisms.(5) Based on the characteristics of the circuit regulation of cancer prognosis. The Embedding Method principles of feature selection, and to build support vector machine model, according to biologists category for GBM survival days to design a support vector machine classifier model based on circuit characteristics GBM cancer prognosis.