MICROBIOTA SIGNATURES OF BLOOD AND TUMOURS WITH METAGENOMIC APPROACHES

Abstract

The human microbiome is a vastly complex ecosystem and influences both the development and treatment of cancer. However, there remains many unknowns about the cancer microbiome and the methods by which it is studied. In this thesis, I examine the microbial profiles of patients from BC Cancer’s Personalized Onco-Genomics project (POG) and National Institute of Health’s Cancer Genome Atlas (TCGA). The tools Kraken2 and MetaPhlAn3 perform differently on low–biomass samples from blood and tumour tissue. Kraken2 misidentifies many human-associated reads as bacteria, leading to false-positive microbial annotations in non-human-associated reads. Using the new Genome Coverage Checker tool, I find true-positive microbial reads. I find no cancer-specific signatures in POG or TCGA samples. Many true–positive taxa are known contaminants. Differences between individuals are likely due to either inherent microbiome differences or contamination during sample preparation. Care must be taken with low-biomass samples so false positives do not influence conclusions.