Talk Title:

Measuring and predicting cancer evolution with genomic data

Abstract:

We combine theoretical population genetics models of evolving populations with machine learning methods to analyze cancer genomic data, with specific focus on whole-genome sequencing. This allows deconvoluting the signal in the data to measure clonal structures, selective advantage coefficients of subclones, and mutation rates, from human tumor samples. The use of a model-based approach also allows to parameterise predictive models to be ‘played forward’ with the aim of anticipating disease evolution.