The closing date for this job has passed; return to the main list for other jobs

Developing copula methods for ABC & applying to disease models

Livestock movements are important for the transmission of infectious diseases, with Foot-and-mouth disease and bovine Tuberculosis (bTB) being important examples. These movements are recorded daily for individual animals and constitute an exceptional record of a dynamic contact network. Bayesian population-based inference methods allow us to fit nonlinear dynamical models to multivariate time-series data, make inferences about disease dynamics and quantitatively compare competing hypotheses regarding underpinning processes. Likelihood-free approaches such as Approximate Bayesian Computation (ABC) [1] allow us to make inferences even on extremely complex models where traditional MCMC methods are not applicable, and such approaches have been used, for example, to successfully describe bTB transmission in GB at the national level [2]. However, the consequences of particular model specifications and ABC metrics are often unclear, posing the question of how well fitted models represent real biological processes, in part due to the complex nature of the multivariate relationships that the model aims to capture. Hence, there is a need for flexible methods to expose and dissect these relationships. Copulas are functions that make it possible to relate multivariate distributions to their one dimensional marginal distribution functions [3] and offer a natural mechanism to model and analyze such relationships.

In this project, supervised across the Roslin Institute and the School of Informatics, the student will develop methods to utilize copulas in ABC approaches to model, study and assess relationships, first working on simple models where reference or ground truth results are known and then using these techniques on real data to uncover fundamental information about disease transmission processes. Applicants with backgrounds in a quantitative subject such as informatics, statistics, mathematics, physics or engineering are encouraged to apply. In the course of the PhD, the student would receive training in mathematical and computational epidemiology, network analysis, machine learning and Bayesian inference techniques. Informal inquiries to

Application procedures
Applications including a statement of interest and full CV with names and addresses (including email addresses) of two academic referees, should be emailed to
When applying for the studentship please state clearly the title of the studentship and the supervisor/s in your covering letter.

PhD position
University of Edinburgh
Closing date
January 16th, 2018
Posted on
December 15th, 2017 14:19
Last updated
December 15th, 2017 16:33