Delineating genome alterations in cancer with long-read and linked-read sequencing

Abstract

Cancer is driven by alterations to the genome. The continued invention and application of new methods has gradually enhanced our ability to characterize of genome alterations in cancer with much greater scale and resolution. In this thesis, I show how the application of emerging long-read and linked-read genome sequencing to the cancer genome enables us to delineate genome alterations that were not previously possible using earlier methods for genome characterization. I will showcase how the application of linked-read genome sequencing to a cohort of renal medullary carcinoma (RMC) samples enable us to identify somatic alterations at the haplotype level and establish the HbS allele as a causal germline allele in RMC. I will further highlight a technical issue we identified with nanopore long-read sequencing in highly repetitive regions such as telomeres, and our efforts to resolve it. Finally, I will demonstrate how the use of long-read sequencing in cancer samples has led to the discovery of neotelomeres and telomere-spanning chromosomal arm fusion events in the cancer genome. Furthermore, by leveraging widely available short-read sequencing data, we inferred the frequency of these two types of genomic events in over 2500 cancer samples across more than 30 cancer types. Together, this thesis highlights the potential of emerging long-read and linked-read genome sequencing in delineating genomic alterations in cancer.