Biomathematics & Statistics Scotland
Microarray technology has enabled scientists to study the expression of thousands of genes in a sample of interest simultaneously (1). Although such gene expression studies have become a standard tool in biological sciences, the high costs of microarrays still put a serious constraint on the number of experiments that can be conducted within a project. The statistical theory of experimental design can be used to optimise the information gained from an experiment (2). However, unlike traditional biological experiments, microarray studies are rarely concerned with a small number of well-defined hypotheses. Rather, there are either a large number of hypotheses or the study is seen as a hypothesis-generating exercise.
In this project we will study the use of sequential strategies, where at each step of the study only a subset of the available arrays is used. The design of the following step then depends on the results obtained from the analysis of the previous one. Such group sequential sampling schemes are well know in medical statistics, where they are often used within clinical trials (3). Sequential methods are potentially cost reducing as they allow to stop an experiment, when enough information has been collected according to some pre-defined criterium. In a microarray context the costs of a hybridisation usually allow little testing of the technology before a study, so a sequential strategy also enables us to pick up potential problems at an early stage. We will also bear in mind that hypothesis testing is only one of several analysis tools. Classification and clustering (4) are other commonly used methods.
Although initial work will use already published microarray data sets, the student will also be involved in the design and analysis of ongoing microarray projects at different institutes, with research topics ranging from obesity and cancer research to host-pathogen interactions of potato viruses.
The student will be based at the Rowett Institute of Nutrition and Health at the University of Aberdeen. Applicants should have good mathematical and statistical knowledge and be interested in the application of statistical methods to complex data sets from biological research.
For further details, contact claus mayer
