Data-driven cross-validation to identify novel therapeutic targets in a diverse range of human diseases.

A common problem in modern biology is combining data, in the form of lists of genes or proteins, from a range of different experiments conducted in different laboratories and under different conditions. In the post-genome era there is an unmet need for a simple, intuitive method to cross-validate and merge different sources of data in order to prioritise targets for future study.
This project will build on a novel method to combine data from disparate sources by systematically evaluating relevant information content in each source. The approach works by iteratively evaluating a weighting factor for each data source. This weighting is determined by how frequently the results of each experiment is replicated in all of the other experiments.
In this cross-disciplinary project you will develop and improve this method, putting into practice some existing plans and your own ideas, and then systematically evaluate the method by comparing it with real and permuted data. The method will be made publicly available as downloadable software from github, and through a web interface hosted by the Roslin Institute.
Concurrently with improving and optimising the method, you will apply it to prioritise gene and protein targets for therapy in a range of applications, including host macrophage responses to life-threatening infection, development of new antibiotics to target specific bacteria, and other applications.
You will work between two world-class research centres at the University of Edinburgh: the Roslin Institute and the School of Informatics. You will be jointly supervised by Ken Baillie (translational genomics, sepsis, intensive care medicine), Michael Gutmann (machine learning, multidimensional data analysis) and Ross Fitzgerald (microbiology, antimicrobial resistance).