Joint Location-Scale Testing in Genetic Association Studies

• Activity Number: 125
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: SSC
• Author(s): David Soave*+ and Lei Sun
• Companies: University of Toronto and University of Toronto
• Keywords: Location-Scale ; Statistical Genetics ; Full Likelihood ; Genetic Interaction

Abstract

The common approach to identifying genetic variant having effects on a quantitative phenotype is to test for phenotypic differences in mean (location) across the three genotypic groups. This approach ignores potential variance (scale) differences due to, for example, interaction with other genetic variants or environmental exposures. In addition, specific choice of measurement scale of the phenotype under study may lessen underlying main effects while creating apparent variance differences that could be harnessed to detect associated variants. We propose a joint location-scale testing framework that includes both a full likelihood approach as well as approaches that combine evidence from the individual location and scale test classes, with the goal of improved variant detection. Efficiency and robustness of the proposed approach are studied via extensive simulation studies and multiple data applications. Complexities of the framework are investigated including effects of phenotypic distribution and frequency of the minor allele on the accuracy of the tests. A generalization of the framework to incorporate group uncertainty commonly encountered in the genetic setting is discussed.