Research
Inferring population history
During my PhD (supervised by Nick Barton), I have used microsatellite data to infer mutation and migration rates and the degree of mating skew in a spatially structured population of Alpine ibex (Capra ibex) in the Swiss Alps, conditionning on known demographic information. In this context, my collaborators and I proposed an approach for exploiting the modular structure of a set of models to make Approximate Bayesian computation more efficient in the face of many parameters.
Collaboration: Mark Beaumont, Iris Biebach, Andreas Futschik, Lukas Keller
Approximate Bayesian computation has become increasingly popular as a method of inference in models for which likelihoods are hard or impossible to compute. The choice of summary statistics is a crucial step and has recently become of interest. My collaborators and I have proposed an approach based on boosting and shown that, in a realistic application, it performs well compared to established methods.
Collaboration: Mark Beaumont, Andreas Futschik
