Situations where there is hybridisation between two different species provide fascinating test cases of the adaptive value of traits that differ among taxa (Barton and Hewitt, 1989). These traits may be related to disease resistance, climate, and more. Hybridisation, particularly anthropogenic, is also significant in a conservation context (Quilodrán et al, 2020). Hybridisation between wildcats and domestic cats in Scotland is a good example (Howard-McCombe et al, 2021). As an initial exemplar, this project will focus on data consisting of whole genome sequences from a large sample of wildcats and domestic cats obtained by our group. The student will extend an existing agent-based model (ABM), implemented in msprime and SLiM (Haller and Messer, 2019) to jointly infer natural selection and demographic parameters using advanced methods in simulation-based inference.
Project Aims and Methods
A major aim is to identify the mode of natural selection on traits that differ between species, and thereby predict, for example, those domestic cat characters that may become more prevalent in the wild population, thereby providing recommendations for conservation actions. However, in addition, it is possible to use this study system to identify potential mechanisms of pathogen resistance, and our group has evidence that this is the case.
Simulation-based inference (Cranmer et al, 2020) is a technique for likelihood-based estimation and for computing Bayesian posterior distributions. State-of-the art methodology, for example that which is implemented in the Python library sbi (Tejero-Cantero, 2020), can achieve this. Investigating how to use high dimensional genomic summary statistics most effectively, such as those generated by Chromosome Painting (Lawson et al. 2012), will be a particularly important component of the project.
More broadly, this project will generate new methodology for carrying out population genomic inference more generally, particularly focusing on hybrid zones. Applications developed during the studentship will be made available for other researchers via a GitHub repository to allow wider application of simulation based inference in evolutionary biology and conservation. Wildcats will be used as exemplars that have the potential to demonstrate conservation impact, though the methodology can be applied in evolutionary biology more generally and suitable data sets from hybrid zones and other evolutionary examples are generally available (e.g. https://datadryad.org/stash/dataset/doi:10.5061/dryad.bnzs7h470).
Candidate requirements
This project will suit students who have an interest in scripting and programming, from a diversity of STEM backgrounds. Interest in Python and R programming would be particularly helpful, and ideally some training in probability and Bayesian inference. We welcome and encourage student applications from under-represented groups. We value a diverse research environment.
Project partners
The supervisory team for this project represents a collaboration between biologists and statisticians with expertise in statistical and mathematical aspects of population biology, particularly related to population genomics and epidemiology. Together, we can provide expert advice on all aspects of the proposed project.
Training
This project will provide training in the use of Python programming and scripting in an HPC environment. For successful candidates with a more biological background, we will also provide training in the necessary mathematical and statistical techniques.
Making an application
When applying for the GW4+DTP Projects in the School of Biological Sciences please choose 'Biological Sciences (PhD)' from the Programme Choice drop down menu (see Prospectus for other Schools offering projects under this DTP). You will need to make a separate application for each project.
It is important that you follow the detailed instructions provided in the How to apply link and read the Admissions Statement in the Prospectus to apply for one of these projects.