Etude de Méthodes Inférentielles et Logiciels pour l'Evolution – EMILE

The acquisition of population genetics data has tremendously improved in the last 20 years. This puts population genetics in the position to gain better understanding about the evolutionary history of populations and species of any kind, but also to propose effective measures for the management of a number of species including crops, domesticated animals, endangered species and species endangering biodiversity or human interests such as bioinvasive and disease-causing species. However, empirical workers do not yet have a well-developed set of statistical tools for realistic inferences from the current wealth of data. The aim of the present project EMILE is to extend current developments in algorithms and methods of inferences in order to provide such tools in the form of user-friendly software. This work will be made possible through the work of twenty participants with complementary background ranging from theoretical statistics to theoretical and empirical population genetics. Specifically, we will extend the range of applications of two classes of algorithms: importance sampling for likelihood-based inferences when feasible, and summary statistics methods based on approximate Bayesian computation (ABC) for more general scenarios. We will root ABC methods in a more rigorous theoretical framework by exploring statistical aspects that have up to now been given little attention such as prediction or model validation. We will address several features of (i) migration (limited dispersal distance, sex-bias, pollen dispersal'), (ii) haploid, partially selfing and partially clonal populations, (iii) different markers type and linkage between these. We will apply these techniques on a diversity of biological systems including plant-damaging viruses and fungi, invasive species that threaten both European biodiversity and European crops. Finally, all methods developed in this project will be distributed by means of free software packages, which should be able to address the following questions: what is the evolutionary history of populations' Which loci are under selective pressure and how strong is selection on these' What are barriers to gene flow' Ultimately, this will hopefully enable their rapid use in contexts with strong economical and societal issues such as reconstructing the sources and history of expansion of the human population, of major bioinvasive species or of plant, animal and human diseases, identifying relevant units of conservation by being able to infer the geographical contours of endangered populations, or estimating the risks of non-OGM plants contamination by neighbour OGM crops through the estimation of pollen dispersal curves in natural populations.