Research On Alternative Splicing And Disease Based On Intelligent Method

Alternative splicing is an essential mechanism in gene regulatory network(CRN).Itdiversify the proteomics through generating two or more isoforms by one gene.Mutation inalternative splicing will lead to the Genetic disease or even cancer. This thesis depends on thecurrent progress in the study on alternative splicing and the related disease. Based on differentkinds of experimental data source,we developed different bioinformatics method correspondto them.This thesis mainly contain these topics：Tissue-specific alternative splicing and disease.The variety of alternative splicing indifferent kinds of tissue is called the tissue-specific alternative splicing. Here we take normaltestis and cancer testis as an example, based on the known EST database to predict thetissue-specific alternative splicing events in normal testis and cancer testis. Furthermore westudy the pattern of the alternative splicing where the pattern refer to nucleotide tendency,chromatin tendency, and the relationship between alternative splicing and disease in testis.Atlast, we make use of the dbEST data in GeneBank, mammal EST and ORF database, andsuccessfully predict the alternative splicing pattern in both normal testis and cancer testis.Species-specific alternative splicing and disease. Alternative splicing in different kindsof species have different patterns, so they have a variety of pathogenesis. We build a frame inorder to study the species-specific alternative splicing and the related disease from a broadview. Firstly, we recognize the conserved pattern in mouse and dog, and try to findcorrespond splicing pattern in different species simultaneously; then we analyze the splicingpattern from many aspects, which include alternative exons, motif, function-based splicingtendency and the distribution of splicing sites. Using the frame we developed, we successfullypredict the conserved pattern of human-specific alternative splicing. We test thehuman-specific alternative splicing and the related disease database using Fisher test. Finallywe found that tissue-specific alternative splicing is tightly connected with disease.Recognizing the RNA-binding protein (RBP) region and predicting the related disease.RNA-binding proteins play an important role in many processes in post-transcription. In orderto find the relationship between RNA-binding proteins’s binding region and the relateddisease, here we provide an HMM based method to predict the binding sites of SFRS1.The proposed method consider the impact of secondary structure, which means we consider boththe sequence information and secondary structure information. Then we use the establishedHHM frame to predict the binding region of RBP. We use the Bayesian network to predict thealternative splicing events involved in RBP binding. Taking cassette exons as an example, wepredict the number of cassette exons involved in the alternative splicing. We also take thedistant between the binding region and the splice site into consideration; finally we search thepoint mutation of RBP binding region in disease database and find the relationship betweenthem. We used SFRS1as a test and found that HMM based method could correctly find thebinding region of RBP and the precision of our method is higher than current methods. In thestudy of cassette exons, we successfully predicted119exons that involved in RBP binding.Furthermore we found that a tight relationship between the RBP binding region and disease.Cancer-specific alternative splicing. Alternative splicing is a direct method to dectetcancer. Some specific alternative splicing may be tightly related to cancer. This thesisanalyzes the cancer-specific alternative splicing using the microarray data source. Firstly, wedesign a method to process the microarray data and detect alternative splicing events; then weestablish a model to predict the level of the motif expression to get the cancer relatedalternative events. Through the experiment on colon cancer and normal colon, we found thatmany colon cancer-specific alternative splicing events, some of which has been annotated.Most of the motif we predict influence the cell tissue, cell stromal, and signal, which maybeinvolved in cell migration and invasion. The result shows that our model can not onlyrecognize alternative splicing events in whole genome using the microarray, but also providea more flexible and reliable method to analyze the gene expression level.