Weighted Alternative Splicing Graphs

Abstract

Alternative splicing of a single pre-mRNA can give rise to di®erent mRNA transcripts. Consequently, alternative splicing is an im- portant mechanism for generating protein di- versity from a single gene. Although alter- native splicing is an important biological pro- cess, standard molecular biology techniques have only identi¯ed several hundred alterna- tive splicing variants and create a bottleneck in terms of experimental validation. In this paper, we propose methods of ob- taining models of weighted alternative splic- ing graphs and ways of generating all alterna- tive splicing forms from a weighted alternative splicing graph. Basically, the method uses the UniGene clusters of human Expressed Se- quence Tags (ESTs) to identify alternative splicing sites. Furthermore, we propose lin- ear time algorithms that correctly produce all possible alternative splicing variants with their corresponding probabilities. Using these methods, we infer several sets of putative al- ternative splicing forms; these results are then compared with methods proposed by others.