|
Abstract:
Alternative splicing (AS) is a process by which the same gene can be used to create different products. The growing understanding of the prevalence of AS and its relation to both genomic diversity and disease has increased the importance of efforts to understand the mechanisms involved in AS. Our group has previously been able to develop both the experimental technology and the matching data analysis algorithms that enable us quantitate thousands of transcripts variants in diverse experimental conditions and tissues. Given this data, the work I will present concentrates on unravelling what we call "the alternative splicing code". The algorithm we developed aims to uncover the underlying combinatorial structure of observed AS levels and explain these in terms of hidden splicing patterns and genomic features. Our generative model accounts for AS measurements, gene expression measurements and RNA sequence motifs, and enables us to optimize both the extraction of AS patterns and the RNA sequence features that relate to them. In the talk, I will discuss some of our current results in this ongoing research and give details of the model and algorithms used to obtain them.