Studies On The Related Issues On The Recognition Of Alternative Splicing

Alternative splicing plays an important role in regulating gene expression. It is widespread in eukaryotic genes. The study of alternative splicing has become one of the most challenging and exciting tasks for genomics and bioinformatics in the post-genomic era. Here, we analyze the recognition of alternative splice events. The dissertation includes three parts as follows:Firstly, we introduce the hidden Markov model (HMM) into the recognition of splice sites. Our method can successfully predict the exact locations of splice sites at the exon-intron junction. The comparison between our method and previously published approaches shows that our method outperforms other methods for the recognition of real and pseudo splice sites in flanking regions of given splice sites. Further more, we used this method to recognize alternative splice sites. Results show that there is no essential distinction between constitutive and alternative splicing in terms of their splice site sequences. Instead, the sequence differences between alternative and constitutive splice sites are of a graded manner with regard to their splicing levels. It suggests that when studying the recognition of alternative splice sites, we should consider splice sites and the influence of splicing regulators, rather than just the splice sites themselves.Secondly, we describe the application of support vector machine (SVM), a machine learning method, for predicting alternative splice events. Starting from the mechanism of alternative splicing, we introduce the competition mechanism of splice sites selection into the prediction of alternative splice sites. This approach allows us to predict alternative splice sites merely based on their genomic sequences. It outperforms an existing method which only considers features extracted from the splice sites themselves. Furthermore, this approach can provide helpful clues to guide further biological experimental analysis and for the study of biogenesis of alternative splicing. We also analyze the sequence features of retained introns, and further use these features to distinguish retained introns from constitutive ones merely based on their own sequences. This method achieves a satisfying prediction performance. Above results indicate that a deep understanding of mechanism governing alternative splicing biogenesis can help us extract effective features for classification.Finally, starting from above results, together with the studies on the identification of other biological sequences, we provide a way for the recognition of biological sequences; that is to say, we should extract features from the biological mechanism of the data in question. In the study of pattern recognition, the key step is to extract features that can represent the essence of data from different classes. For the pattern recognition in bioinformatics, the biological mechanism might be the right start point.