Measuring tumour evolution at the genetic and epigenetic level in individual patients from cancer genomic data

Abstract

High-throughput genomic data from cancers uncovered high intra and inter-tumour heterogeneity and subclonal architecture of cancer cell populations. If we consider cells as asexually reproducing individuals, we can apply evolutionary theory to cancer, as all three building blocks of evolutionary dynamics - replication, selection, and mutation - are also the defining characters of cancer development. Studying cancer evolution in humans is imperative to predict the course of the disease and develop better therapeutic strategies. Currently, there is a lack of mathematical and computational models that describe the effects of clonal selection in cancer data for growing populations. One of the reasons being that selection depends on many factors, such as fitness, context, and spatial constraints. In this thesis, we developed a stochastic simulation model of spatial tumour growth from which we can generate the genomic data we expect under different conditions and sampling methods. The model enabled us to monitor the effects of sampling bias on cancer genomic data as well as the effects of spatial constraints on tumour growth dynamics. In the second part of the thesis, we also tried to study the links between genetics and epigenetics that influence cancer formation and progression. We developed methods to test our hypothesis that different mutational processes (giving rise to distinct mutational signatures) are active in epigenetically different regions of the genome, and a model that infers times of different chromatin aberration events. Overall, this thesis shows the importance of coupling mathematical and computational modelling with experiments to gain a better understanding of cancer initiation and progression and consequently achieve better clinical performance.