Identification and Registration of Hits in Genome-Wide Association Studies

Supervisor: Lloyd Elliott

Multivariate genome-wide association studies (GWAS) provide a vector function of the genome (the elements of the vector are the p-values of each phenotype). Typically, these objects are examined by hand in order to determine "hits" (local minimum). However, when the number of phenotypes is large, an automation of this process is required. The project involves 1) implementation of a method for determining "hits", 2) extension of the method to allow registration of hits between phenotypes (i.e., determine when two hits on two phenotypes are close enough to be called the same location.) 3) development of Bayesian nonparametric theory for this problem.