Simon Tavaré - The genealogy of branching processes and the reconstruction of tumor histories

Genealogical approaches have found many applications in population genetics, where the coalescent has played a crucial role in understanding molecular variability in natural populations. Related methods are also proving useful for the study of mitotic division processes. We have been using a quantitative analysis of microsatellite mutations in colorectal tumors to infer their histories, and to shed light on the likely nature of the adenoma-carcinoma sequence. The main tool we use is an algorithm for generating the genealogy of a sample of cells evolving according to a discrete time (possibly non-Markovian, time-inhomogeneous) branching process. A number of applications, including estimation of the age of a tumor and generalizations to multitype processes, will be discussed.