The Taub Faculty of Computer Science Events and Talks
Micha Sammeth (Genomic Regulation, Barcelona, Spain)
Sunday, 15.06.2008, 13:30
Eukaryotic splicing structures are known to involve a high degree of alternative forms derived from a premature transcript by alternative splicing (AS). With the advent of new sequencing technologies, evidence for new splice forms becomes more and more easily availablebit by bit revealing that the true splicing diversity of "AS events often comprises more than two alternatives and therefore cannot be sufficiently described by pairwise comparisons as conducted in analyzes hitherto. Further challenges emerge from the richness of data (millions of transcripts) and artifacts introduced during the technical process of obtaining transcript sequences (noise)especially when dealing with single-read sequences known as expressed sequence tags (ESTs). We describe a novel method to efficiently predict AS events in different resolutions (i.e., dimensions) from transcript annotations that allows for combination of fragmented EST data with full-length cDNAs and can cope with large datasets containing noise. Applying this method to estimate the real complexity of alternative splicing, we found in human thousands of novel AS events that either have been disregarded or mischaracterized in earlier works. In fact, the majority of exons that are observed as mutually exclusive in pairwise comparisons truly involve at least one other alternative splice form that disagrees with their mutual exclusion. We identified four major classes that contain such optional neighboring exons and show that they clearly differ from each other in characteristics, especially in the length distribution of the middle intron.