Based On Gene Expression Data To Predict Transcriptional Regulatory Factors And Transcriptional Regulatory Elements Active

As the most powerful high-throughput technology in automatic gene expression level measurement, microarray enables researchers to study thousands of genes simultaneously in whole genomic scale. It has great impact on both fundamental researches and clinical application to find how to infer reliable regulatory relations from the'data ocean'with the combination of bioinformatics and data mining tools. In this study, we perform detailed analysis on the inference of activities of transcription factors (TF) and transcriptional binding elements based on the integration of gene expression data. The major innovations of this research are: extending the application of TF activity from S.cerevisiae biological processes to human cancer researches; proposing a novel method to inferring the activities of transcriptional regulatory element; suggesting that between S.cerevisiae and S.pombe., the cis-elements associated with cell cycle regulation are highly conserved, whereas individually the presence of an element in the promoter of a specific gene is dynamic.The first chapter introduces the background of the subject, including the basic steps of microarray data analyses and the general methods on the researches for gene regulation.The second chapter introduces a published method for inferring transcription factor activity, which named BASE (Binding Association with Sorted Expression). We applied this method to five prostate cancer microarray datasets and identified a few transcription factors related to prostate cancer. Some of TFs have been validated by published literatures. Some novel TFs, such as proteins in Homobox superfamily and bHLH family, may provide clues to disclose the mechanism for the progression of prostate cancer.The third chapter introduces a novel method for the identification of cell cycle regulatory elements(CCRE) by examining the periodicity of element activity profiles inferred from the expression profiles of their target genes. To validate this method, we applied it to a S.cerevisiae cell cycle microarray data. The results show that our method can detect 7 of 8 well known cell cycle regulatory elements. Furthermore, we compared our method with a sophisticated method. It is indicated that the elements identified by our method have stronger associations with cell cycle process, suggesting that our method is an effective method for the identification of CCREs.The four chapter focuses on the application of the inference of CCRE. We identified the CCREs of S.cerevisiae and S.pombe. based on published cell cycle microarray datasets and then systematically compared these CCREs between these two yeast species. We find that the majority of CCREs are conserved in both species and their abilities to regulate periodically expressed genes are maintained. Despite the conservation, these CCREs have tremendously high turn-over rates in the promoters of target genes and exhibited quite distinct peak times in their activities between S.cerevisiae and S.pombe.The fifth chapter makes a prospect on the future development of inferring activity of transcriptional regulatory components.