| Pre-mRNA splicing is a fundamental step in eukaryotic gene expression. It remains a mystery how exons are recognized among introns by the splicing machinery, especially in mammalian genes. My doctoral study has been focusing on identification and verification of cis splicing signals that facilitate exon recognition in the human transcriptome.; Recent advances in the field of splicing have revealed that splicing signals are degenerate but abundant.; In this thesis, I will discuss my efforts to identify splicing motifs in exons as well as introns. The thesis will be organized into 8 chapters and an appendix. In the first chapter, I will introduce some background knowledge and recent advances in the field of splicing. In Chapter 2, I will present the identification of several groups of putative intronic splicing signals by a state-of-the-art machine learning technology, Support Vector Machines (SVMs). The success in finding these motifs indicated that the intronic flanking regions (designated as "flanks" hereafter) of exons are probably playing a general role in splicing. Chapter 3 extends Chapter 2, describing the discovery of the dichotomy of intronic splicing signals. In this chapter, I will also present our global analyses of secondary structures in flanks and some genetic experiments that highlight the general importance of flanks to splicing. In Chapter 4, I will switch to exonic splicing signals, discussing the computational definition of putative splicing enhancers and silencers. Representatives of the identified enhancers and silencers have been tested in vivo. Chapter 5 shows some further tests on the sequence motifs we identified in Chapter 4. We used site-directed mutagenesis to knock out predicted enhancers in natural exons and assayed the splicing phenotype corresponding to the mutations. These experiments not only confirmed the validity of most sequences in our list but also pointed out striking non-redundancy among multiple enhancer elements within a same exon. Chapter 6 is a genomic analysis on the relationship between stop codons and splicing. It provides evidence against a general role of translatability of exons in splicing. In Chapter 7, I will discuss the development of a computational algorithm, CellView, which makes use of all the putative splicing signals identified in the preceding chapters and mimics exon recognition in silico. In Chapter 8, I will summarize, introduce some ongoing projects and discuss future directions. (Abstract shortened by UMI.)... |
