WAVELET BASED FUNCTIONAL MODELS FOR TRANSCRIPTOME ANALYSIS WITH TILING ARRAYS

For a better understanding of the biology of an organism a complete description is needed of all regions of the genome that are actively transcribed. Tiling arrays can be used for this purpose. Such arrays allow the discovery of novel transcripts and the assessment of differential expression between two or more experimental conditions such as genotype, treatment, tissue, etc. Much of the initial methodological efforts were designed for transcript discovery, while more recent developments also focus on differential expression. To our knowledge no methods for tiling arrays are described in the literature that can both assess transcript discovery and identify differentially expressed transcripts, simultaneously. The wavelet based functional model developed in this paper is designed to fill this methodological void. As opposed to existing methods, our statistical framework also permits a natural integration of preprocessing into the standard statistical analysis flow of tiling array data. We use Johnson transformations, which are based on cumulants, for computing false discovery rates (FDRs) and Bayesian credibility intervals for the estimates of the effect functions within the data space. A case study illustrates that our model is well suited for a simultaneous assessment of transcript discovery and differential expression, while remaining competitive with methods that perform only one of these tasks