Computational analysis of transcriptional and post-transcriptional regulation of gene expression

The regulation of gene expression is fundamental to all life on Earth. Dynamic but precise control is vital to cell survival and function, and takes place at various tightly interwoven levels. In this thesis, we review and study the crosstalk between different types of regulators, including epigenetic regulators, transcription factors (TFs), RNA-binding proteins (RBPs) and microRNAs (miRNAs). First, we focus on the interplay between miRNAs and other types of regulators, in particular TFs and epigenetic regulators, both of which are strongly enriched among the predicted targets of miRNAs. Indeed, the direct interplay of miRNAs with other regulators that have genome-wide impact is one possible explanation for the reported importance of miRNAs to fundamental biological processes, including cell fate. We introduce a computational strategy that we apply in order to infer the transcription regulatory circuitries that act downstream of embryonic miRNAs. More precisely, we analyze genome-wide expression changes with an extended motif activity response analysis (MARA) model in order to identify transcriptional regulators that are direct targets of embryonic miRNAs and change in activity upon expression of the miRNAs. We experimentally validate our most promising predictions and integrate the extended MARA model into an automated system in order to make it available to other researchers. We demonstrate its application by modeling diverse high-throughput datasets, including paired liver biopsies of patients with chronic hepatitis C virus infections. Finally, we study alternative cleavage and polyadenylation, a process that impacts gene expression in various ways, including modulating the presence of cis-regulatory elements, such as miRNA and RBP binding sites, which tend to be located at the 3' ends of transcripts. We demonstrate that global shortening of untranslated transcript regions, which is associated with proliferative states, has a very limited effect on mRNA stability and protein output. By analyzing a large array of high-throughput 3' end sequencing data, we create comprehensive catalogs of 3' end processing sites for both human and mouse. Moreover, we identify novel cis-regulatory motifs that are involved in cleavage and polyadenylation, and point out a regulator, HNRNPC, that binds to one of the motifs, thereby globally impacting the usage of cleavage and polyadenylation sites.