Inference of gene regulatory interactions from deep sequencing data

This thesis describes our work on the inference of gene regulatory interactions from deep sequencing data. In the first part we start with an introduction to the problem of inferring transcription regulatory interactions (Chapter 2). Then we describe detailed methods of promoterome construction in human and mouse genomes from deep CAGE data (Chapter 3), and show how promoters together with gene expression data can be used to infer transcription regulatory interactions (Chapter 4). Finally, we show an application of this methodology to a human macrophage lineage undergoing differentiation accompanied by a detailed experimental validation of the predicted network structure (Chapter 5). Work presented in Chapter 5 comes from the FANTOM4 project.
In the second part we focus on the regulatory functions of two small nucleolar RNAs (snoRNAs). In Chapter 8 we describe an atypical function of snoRNAs - regulation of the mRNA alternative splicing process by a particular class of mouse snoRNAs (MBII-52 variants). These are of great interest, as the locus encoding MBII-52 is linked to Prader-Willi Syndrome. Methods include in silico RNA hybridization screens and experimental confirmation of the predictions. In Chapter 9, we report a discovery of the first virus-encoded snoRNA in Epstein-Barr Virus (EBV). Again, we show that the function of this snoRNA (v-snoRNA1) is atypical: it is processed into small, 24 nt long fragments that can function as microRNAs.