Detection of Associations Between a Set of Rare Variants and Multiple Continuous Phenotypes

• Activity Number: 37
• Type: Contributed
• Date/Time: Sunday, July 31, 2016 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Zhonghua Liu and Xihong Lin and Zilin Li*
• Companies: Harvard and Harvard T.H. Chan School of Public Health and Harvard T.H. Chan School of Public Health
• Keywords: Multiple Phenotypes ; Multivariate linear mixed model ; Rare variants ; DNA Sequencing ; Variance component test

Abstract

We consider in this paper jointly testing the associations between a set of rare variants in a genomic region and multiple continuous phenotypes. Since the correlation structure between multiple continuous phenotypes could be arbitrary, and the association patterns with rare variants also vary across the whole genome, we therefore propose novel tests based on multivariate linear mixed models by assuming the genetic effects of rare variants on multiple phenotypes follow an arbitrary distribution with a common mean and a pre-defined covariance matrix. This model allows the accommodation of both a fixed effect and random effects. We derive a score test for the fixed effect and a modified variance component score test for random effects, so that these two score testing statistics are independent. We propose several methods to combine these two independent score tests. The p-values of the proposed tests can be computed analytically. We conducted extensive simulation studies to evaluate the performance of our tests under practical settings. An application to the five metabolic syndrome quantitative traits sequencing association study demonstrates the usefulness of our methods.