| Alternative splicing plays an important role in processes such as development, differentiation, and cancer and is tightly regulated in a tissue-specific manner. The most recent estimate of its prevalence in human was as high as 65%. The much higher prevalence makes it an imperative to produce a comprehensive catalog of all splice variants and develop a better understanding of its regulation and function. In this dissertation, we described the computational algorithms we developed that address the identification of tissue-specific alternative splicing regulatory elements, the prediction of tissue-specific alternative splicing events, and the analysis of the functional consequences of alternative splicing.; The most important contribution of this dissertation is the method we developed to train tissue-specific alternative splicing classifiers using tissue-specific regulatory elements and splice site features. It is the first such method we know of to predict alternative splicing events in a tissue-specific manner. The tissue-specific classifier trained on normal nervous system specific cassette exons achieved an ROC score of 0.857. 50% of the top genome-level predictions made by another classifier for more than 14,000 exons were found to be alternatively spliced with support of full-length transcripts. Such classifiers are very useful for biologists interested in investigating tissue-specific alternative splicing events by helping to direct and prioritize their efforts and resources. In addition, they can make valuable contributions toward the creation of a catalogue of all splice variants and their respective tissue distributions.; We also identified a number of protein domains disproportionately distributed among alternatively spliced genes and also those that tend to be spliced-out. The proteins having these domains are predominantly involved in the processes of cell communication, signaling, development and apoptosis. They function mostly as enzymes, signal transducers, and receptors. These findings suggest that alternative splicing is an important information processing step. Somewhat surprisingly, 28% of all occurrences of spliced-out domains are not performed by straightforward exclusion of exons coding for the domains but by inclusion or exclusion of other exons to shift the reading frame while retaining the exons coding for the domains in the final transcripts. |
