Considerations for the genome-wide profiling of colorectal neoplasia

Download files
Access & Terms of Use
open access
Copyright: Zarzour, Peter
Altmetric
Abstract
Recent technological advances have made possible the generation of whole genome sequences at high throughput and low cost. Recent studies have capitalised on these advances to profile different types of colorectal tumours. The statistical framework currently used in such analyses requires multiple biological replicates with corresponding normal bowel samples to define average tumour-specific changes. A limitation of this approach is that it cannot be used to interpret data from a single patient sample, and this must be resolved if genome-wide profiling is to be clinically useful. To this end, this thesis explores analytical approaches for genome-wide profiling to detect tumour-specific changes from single patient samples and segregate colorectal tumours. Genetic changes were first considered: SNP arrays from colorectal adenomas and stage I and III carcinomas were used to identify copy number alterations from individual samples that could distinguish tumour stages. The major impediment to such analysis was the biological variability between tumour samples and the lack of genomic aberrations in subtypes of colorectal neoplasia. Next, RNA-seq-based expression data was explored to improve stratification sensitivity, and to establish an analytical approach to eliminate the requirement for technical and biological replicates and paired normals. Using data from 16 normal human colonic mucosa samples, a gene percentile rank approach was employed that successfully circumvented technical (read alignment, sequence mappability, normalisation and sample preparation) and biological variability to identify a robust normal colonic expression profile. As proof-of-principle, this gene-rank approach was tested for its ability to correctly stratify 16 adenomas into 3 morphological subtypes. The rank approach was found to be superior to single sample differential expression and tumour-normal fold change. The gene-rank approach was externally validated by identifying gene expression signatures in freshly isolated stem and differentiated cells from the colonic epithelium of 44 Lgr5-GFP mice. By investigating publically-available RNA-seq datasets, a “stem cell profile” that was characterised by the upregulation of stem-associated genes and downregulation of differentiation-associated genes was identified in 4.4 % of 562 CRCs. The analytical approaches described herein offer new insights into the genome-wide profiling of colorectal neoplasms from single patient samples.
Persistent link to this record
Link to Publisher Version
Link to Open Access Version
Additional Link
Author(s)
Zarzour, Peter
Supervisor(s)
Ward, Robyn
Hesson, Luke
Creator(s)
Editor(s)
Translator(s)
Curator(s)
Designer(s)
Arranger(s)
Composer(s)
Recordist(s)
Conference Proceedings Editor(s)
Other Contributor(s)
Corporate/Industry Contributor(s)
Publication Year
2015
Resource Type
Thesis
Degree Type
PhD Doctorate
UNSW Faculty
Files
download public version.pdf 27.93 MB Adobe Portable Document Format
Related dataset(s)