Open Collections

Chu, Ching Pan

University of British Columbia

Co-expression analysis is a popular method for probing regulatory circuitry that governs gene expression. Despite many thousands of such studies, the insight gained into regulation has been limited. In Chapter 1, I provide a literature review on the challenges associated with current approaches to co-expression analysis. My thesis is based on the hypothesis that co-expression computed across bulk tissue samples will struggle to identify intracellular dynamic regulation due to unwanted sources of co-variation among genes. In Chapter 2, I assembled a set of experimentally validated transcriptional regulatory interactions by manually curating the research literature, in order to enable evaluation of co-expression networks. I developed a curation protocol for recording reports of low-throughput experimental evidence with detailed information. With the help of a team of curators, I assembled a collection of 1,499 transcription factor and target interactions, which has been expanded to 4,387 interactions since the original publication. In Chapter 3, I studied the propagation of co-expression signals from single cells to bulk tissue samples using simulations. I developed a novel single cell expression data simulator to generate synthetic bulk samples with fully resolved cellular makeup. I show that there are highly specific and unlikely conditions under which regulatory signals may be preserved at the bulk tissue level. These findings shed light on why regulatory network inference using co-expression analysis has proved challenging. In Chapter 4, I analyzed real single cell and bulk tissue datasets to empirically compare co-expression patterns at different levels of cellular resolution. I found co-expression at different levels was largely discordant, validating the theoretical predictions from the previous chapter. Importantly, by using a simple aggregation strategy, I constructed cell type specific co-expression networks that capture reproducible signals reflecting dynamic regulation. In Chapter 5, I conclude my thesis by summarizing its implications for the larger field of co-expression research. A key contribution of this thesis is the explicit clarification of the aims of co-expression analysis and an evaluation of its utility given those aims. Specifically, I posit that co-expressions at the single cell level within cell types is most likely to carry signals of intracellular dynamic transcriptional regulation.

Text

2024-08-21

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/88984

Bioinformatics

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/