Developing computational methods to enhance understanding in glioma progression

Abstract

Longitudinal sampling has provided a wealth of information and opportunities towards studying cancer evolution and heterogeneity. This has helped to determine the origin of some cancers, identify evolutionary patterns and decide how to best treat and study some cancer types. This thesis first presents a longitudinal study of patient-matched initial and recurrent glioblastoma samples. This gives insight to typical analysis as well as some of the challenges that occur through this type of sample collection. The latter part of the thesis presents an alignment-free method as an opportunity to quickly obtain an overview of pre-alignment sequencing data. Alignment-free sequence comparison is a method which has been in development in a wider biological context since the 1970s. However, thus far it has only been used in a phylogentic sense to investigate relationships between organisms. This thesis explores the possibility of repurposing these methods so that they can explore relationships between multiple samples taken from an individual with a cancer diagnosis. This required the development of a bespoke software, given the additional complexities of the genomic landscape of cancer, followed by an in-depth study of the appropriate parameters to be applied. Finally, the utility of the software was assessed in the context of genomic cohorts generated from longitudinal sampling of cancer in a single patient for three patients with glioma and three patients with clear cell renal cell cancer. Of the six patients evaluated, two produced trees with the same topologies as those produced using traditional, alignment-based methods, two displayed minor variations and two displayed larger variations in topology. These results indicate that alignment-free methods for sequence comparison can be a useful exploratory tool in investigating the relationship between tumour samples within a single patient. Further investigation is required to determine the cause of this variation, whether it represents noise in the data or genetic variability not accounted for when using traditional methods.

Thesis embargoed until 31 July 2023.

Date of Award	Jul 2022
Original language	English
Awarding Institution	Queen's University Belfast
Sponsors	Cancer Research UK
Supervisor	Darragh McArt (Supervisor), Kevin Prise (Supervisor) & Manuel Salto-Tellez (Supervisor)