Study On The Characters Involved In Splicing Mechanism Of Eukaryotic Genes

Spliced structures (the exons are intervened by the introns) are extensively existed in eukaryotic genes. Pre-mRNA splicing processing after primary transcription is the important mechanism for gene expressing regulation. At the same time, alternative splicing is one of the important reasons responding to the complexity of transcriptome and plays important role in development, tissue specific expression of genes, disease, etc. It is thus of great interests to study the splicing mechanism either by bioinformatics or by experimental approaches.The immense experimental data implicate much information related with splicing mechanism. Based on available experimental data, this thesis aims to reveal the related characters of splicing mechanism. Features of splicing structure are studied on genome scale annotation data. The cis-elements involved in tissue specific alternative splicing are discovered using word contrast algorithm. EIparser, a splicing alignment algorithm, and Expath, an alternative splicing analysis method based on graph theory, are developed in this thesis. In addition, a standard dataset representing the transcriptome and a system for evaluating alternative splicing are also developed using the standard genomic DNA sequences and cDNA sequences based on high performance computing techniques.The research on features of gene split structure takes use of the genome annotation data with experiment evidences to analyze the characters of length and splicing signals, which are widely used in various gene prediction methods. The phenomenon that the length of the first introns is especially long has been discovered and maybe relate with their special functions. We especially analyze the stop codon signals in all three possible open reading frames. The result indicates that both the untranslated frames in spliced structures have sensitive supervisor mechanism for avoiding transcript mistakes. The alternative splicing sites have weaker site signal strength under the WAM than constitute splicing site, but both of them have overlapped distribution with flase site. The new SVM method with features of site base content, tri-tuple content, signal strength and local optimal site pair has better ability to classify true and false sites than the method only based on signal strength.The research on cis-element of alternative splicing takes use of the dataset of brain specific splicing sites in ASAP. Based on the discovery that sites are defined by the competition among the splicing factors, the dataset have been classified according to different splicing forms. A word contrast algorithm is introduced to build a set of over-representation oligonucleotide elements contrast against the control dataset. Part of the elements match with the known splicing regulated sequences. The evolutional conservation studies are performed for the sequences have computed elements. The results indicate that thest elements have potentially regular function. The elements set thus could stand for a good chance to be real elements. The relation of regular elements and alternative splicing forms is discussed based on the elements with known function.EIparser, the splicing alignment method, takes use of the study result of splicing structure features discovers more appropriate splicing structures than other methods and expresses the results in a sense of biology. EIparser has been used in the construction of standard transcripts dataset based on various RefSeq databases of model species. Moreover, Expath, the algorithm of alternative splicing analysis based on graph theory and genome data, has better computer performance and reliable results than traditional transcripts data analysis methods. The system of transcriptome alternative splicing analysis based on EIparser, Expath and high performance computer has been constructed.