The Epigenome through the Lens of the Transcriptome and How it Regulates Tumour Development in Cancer
Year: 2020
Type: dissertation
Abstract: Battling complex human disease requires deep understanding of the disease etiology at the molecular level. Recent international collaborations provide a multitude of datasets assessing changes in DNA at the single-nucleotide resolution, chemical modifications such as cytosine methylation, protein binding to DNA such as transcription factors (TFs), and gene expression for a variety of patient samples and cell lines. Getting the most out of these datasets, however, requires methods which address technical difficulties and provide valuable biological insights. Assessing cytosine methylation through sequencing or arrays, to date, remains the most widely used assay for investigating the epigenome. The experimental procedure, however, changes the nature of the DNA assembly that the sequencing reads could map to. In this dissertation, I will propose a solution and provide publicly available datasets to allow researchers use DNA methylation assessments properly. Identifying TF binding sites remains an important challenge in epigenomics. The available assays, such as ChIP-seq, however, require a large amount of cells not always available from patient tissues. For decades, researchers have relied on the sequence preference of TFs instead of directly assessing TF binding. In this dissertation, I will show that relying solely on TF sequence preference will bring many false negative findings as well as previously reported false positive findings (futility theorem). I propose a solution, Virtual ChIP-seq. To predict TF binding, Virtual ChIP-seq uses as input the data of two assays which require only minimal tissue: RNA-seq and ATAC-seq. These assays assess the transcriptome and the accessible chromatin, respectively. For certain TFs, Virtual ChIP-seq predicts cell-type–specific DNA binding with high accuracy. I provide predictions of Virtual ChIP-seq for binding of 36 TFs in 31 human tissue types as a public resource. In the last part of this dissertation, I investigate the transcriptome and epigenome in a specific group of tumours: squamous cell carcinomas infected with human papillomavirus (HPV). I found that HPV integration into the host genome modifies the local host epigenome and transcriptome, possibly because of a conserved CTCF binding site within its 8 kbp genome. This dissertation introduces publicly available tools that allow the research community better assess the epigenome and it also enhances our understanding of how HPV contributes to tumourigenesis.
Source: TSpace (University of Toronto)
Author: Mehran Karimzadeh Reghbati
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W3139828953