 |
 |
The genetic epidemiology group at the University of Leicester is offering the following MRC-funded PhD project to work with Professors John Thompson and Paul Burton on the following topic: Modelling non-random ascertainment in genetic family data Brief Description: Studies of the genetics of complex diseases, such as asthma and ischaemic heart disease, often rely on recruiting families of subjects. It is rarely possible to recruit families at random, however other methods of sampling can lead to a severe form of selection bias known as ascertainment bias. For instance, ascertainment bias arises if families are only sampled when at least one member is affected and so at-risk families with no affected members are unobserved. Statistical models for addressing the problems of ascertainment bias involve high-dimensional integration and are very difficult to implement fully even using the latest Markov Chain Monte Carlo (MCMC) methods. To compound the problem, published work from the Genetic Epidemiology Group in Leicester has demonstrated that some modified MCMC approaches that are practical, do not produce full adjustment for the bias. This PhD project would build on this work in order to understand better why practical MCMC methods sometimes fail and then to develop improved methods for adjusting for the ascertainment bias. The problem will be tackled from two complementary directions: (a) fitting conventional models by maximum likelihood or MCMC methods that simulate pseudo-observations to reflect the unobserved families; (b) post-processing the estimated parameters from approximate MCMC analyses to improve the adjustment for ascertainment bias. Real and simulated data sets will be used to assess the performance of the various approaches. The stipend is the usual £12,000 tax-free per annum. Knowledge of genetics is not essential if the candidate has a good background in Mathematics/Statistics and is interested in biological applications. The topic is also suitable for a numerate biologist who perhaps has a Master's Degree with a substantial statistics component. However, these are only rough guidelines and we would encourage anyone who is at all interested to contact John Thompson at [log in to unmask] Nuala A. Sheehan, Department of Epidemiology and Public Health, University of Leicester, 22-28 Princess Road West, Leicester LE1 6TP, UK Tel: +44 (0)116 2525456 Fax: +44 (0)116 2523272 Email: [log in to unmask]