Weighted Pseudolikelihood for Analysis of Multiple Secondary Outcomes in Genetic Association Studies

• Activity Number: 649
• Type: Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: ENAR
• Author(s): Elizabeth Schifano*+ and Tamar Sofer and David C. Christiani and Xihong Lin
• Companies: University of Connecticut and Harvard School of Public Health and Harvard School of Public Health and Harvard School of Public Health
• Keywords: Ascertainment ; Model selection ; Multivariate ; variance component test ; weighted BIC

Abstract

There is increasing interest in the joint analysis of multiple outcomes in genome-wide association studies (GWAS), especially for analysis of multiple secondary outcomes in case-control studies. By taking advantage of correlation, both across outcomes and across Single Nucleotide Polymorphisms (SNPs), one could potentially gain statistical power. We propose novel statistical testing and variable selection procedures using pseudolikelihoods to identify SNP sets (e.g., SNPs within a gene), as well as individual SNPs, associated with multiple outcomes. For multiple secondary outcomes, we use a weighted pseudolikelihood approach to account for case-control ascertainment in testing and variable selection, and additionally propose a weighted Bayesian Information Criterion for tuning parameter selection. We demonstrate the effectiveness of both procedures through theoretical and empirical analysis, as well as in application to investigate SNP associations with smoking behavior measured using multiple secondary smoking outcomes in a lung cancer case-control GWAS.