Modelling the 3D-Genome: The development of network theory approaches to characterise and predict active enhancers

Abstract

Gene regulation is an important mechanism that ensures the correct functioning of a cell and is generally orchestrated by gene regulatory elements such as transcriptional enhancers. Identification of these genomic regions are important in understanding a wide range of phenomena such as evolution, homeostasis and disease. During gene regulation, signals pertaining to transcriptional activation are transferred across the chromatin regulatory network from enhancers to genes in the form of transcription factors and cofactors that in turn, recruit transcriptional machinery such as RNA Polymerase II to increase the rate of gene transcription. Conceptually, we describe this as a flow of information from enhancers to genes, mediated by the chromatin conformation. We exploit this relationship in order to decode the regulatory landscape of genes and identify active enhancers. This thesis outlines the difficulties associated with identifying pathogenic mutations in the non-coding genome due to a lack of robust enhancer annotations. We use network theory to annotate these regions and develop a new method, 3D-SearchE, that serves to predict the location of novel putative active enhancers. 3D-SearchE achieves this by reverse engineering the flow of information between enhancers and genes to calculate an imputed activity score (IAS) at intergenic loci. We show that intergenic loci with a high IAS are also present for other enhancer associated features including the histone marks H3K27ac, H3K4me1 and H3K4me2, P300, CAGE-seq, Starr-seq, eQTLs and RNA Polymerase II. 3D-SearchE successfully leverages and summarises the relationship between the 3D organisation of chromatin and global gene expression and represents a novel enhancer associated feature that can be used to predict active enhancers.