Research On Prediction And Analysis Of Transcriptional Regulation And Construction Of Regulatory Networks Based On High-throughput Sequencing Data

Available biological data grow rapidly with the wide application ofhigh-throughput sequencing technology. Interpreting the biological data anduncovering biological knowledge behind the data have increasingly become thefocus of research in bioinformatics. Transcriptional regulation is the most basic andimportant regulatory mechanism, prediction and analysis of transcriptionalregulation and interaction of transcriptional regulatory networks in the humangenome could deepen understanding the essence of the transcription process whichrevealing human disease pathogenesis.This study, based on high-throughput data generated by next-generationsequencing technology, modeled for transcription factors that regulate rules on thelevel of hereditary, designed and developed computational models and analysismethods to predict transcriptional regulation and transcription factor regulatorynetworks. Besides, this study discussed separately regularization method, selectionof prediction measures and significant threshold selection method, developed thetheoretical model and applications for high-throughput data. The main contentincludes:(1) A normalization method based on genomic functional information wasproposed.To overcome the weakness of current high-throughput sequencing datanormalization method without take into account the impact of structural genomicsfor biological data distribution, this article proposed a new LOWESS normalizationmethod based on functional annotation of the genome. This method according todifferences in biological function or different research purposes normalized genomedata based on its functional annotation. Compared with traditional methods, thismethod has higher specificity and flexibility and lower the time and spacecomplexity.(2) PolII was adopted as new data source, a prediction method of transcriptionalregulation was proposed and transcription factors regulatory networks wereconstructed.RNA polymerase II (PolII) binding data on the genome was adopted as newpredictor data source of transcriptional regulation to construct a highly accurateprediction workflow and transcription factors regulatory networks. Comparativeanalysis of predictions between PolII data and gene expression data following thesame workflow proved PolII data can provide richer information for the predictionof transcriptional regulation. This study also proposed a new integrated scoring SIS significant measure and the corresponding significance threshold determined criteriato overcome the weakness of existing methods of deduced cooperative relationsbetween TFs. Based on the literature and experimental data confirmed thecorrectness of the identification of transcriptional regulation of TFs and theirregulatory networks.(3) Analytical method was proposed for multiple types of regulatory factorsinvolved in the transcriptional regulation and transcriptional regulatory networkswith regulatory loops were constructed.For multi-type factors involved in the transcriptional regulation, a model forTFs-microRNA regulatory regions was constructed, TFs and co-regulators involvedin the regulation of were predicted and target genes were predicted. microRNAtarget genes, regulatory TFs, microRNA and forward and feedback loops built basedon identified regulatory structure were constructed transcriptional regulatorynetworks with regulatory loops in HeLa cells. These new identified regulatory modefurther increase understandings of regulation mechanism of TFs in tumor.(4) A prediction method of transcriptional regulation about TF-lncRNA wasestablished.A native Bayesian model based on integration of multi-data was proposed topredict the TFs regulation for lncRNA. The model based on sequence information,chromatin status information, epigenetic information and other data sources ofinformation, through the construction of priori data mode and parameterestimation to predict the posterior probability of TF-lncRNA regulation.Comparement with ChIP-Seq experiments of corresponding TFs proved theprediction method could make an accurate prediction of TFs’ regulation with orwithout corresponding feature data of TFs.