Inference of Alternative Splicing from Tiling Array Data

Abstract

Alternative splicing (AS) is an important mechanism implicated in eukaryotic gene expression, whereby exon segments of precursor-mRNA transcripts are joined together in different arrangements corresponding to diverse isoforms of mature mRNA. Accumulating evidence suggests that in many instances this process is specifically regulated and contributes to the structural and functional diversification of tissues and cell types. Furthermore, several studies support the view that environmental stresses dramatically impact on AS and reported the presence of novel transcript isoforms in response to biotic or abiotic stresses. Since specific regulation of AS in plants is a largely unexplored field of research, large-scale approaches aimed at monitoring AS on a genome-wide level are of increasing importance to gain insights into tissue-specific splicing regulation and to study the effects of changed environmental conditions on pre-mRNA splicing.Here, we describe the concepts of a traditional statistical approach, and a more recently developed machine learning-based method for AS detection from tiling arrays. The here presented approaches were employed for the detection and profiling of AS events in the model plant A. thaliana, and applied to a large dataset comprising transcriptomic expression data from 11 tissues and 13 stress conditions.