A General Meta-Analysis Approach for Haplotype Association Results in Family and Unrelated Samples

• Activity Number: 84
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Shuai Wang*+ and Josée Dupuis
• Companies: Boston University and Boston University School of Public Health
• Keywords: meta-analysis ; haplotype ; association test ; family samples ; unobserved/missing data ; generalized least square

Abstract

In genetic association studies, meta-analysis has been widely used to improve power to detect signals. Some approaches were developed and successfully applied to meta-analyze association tests of single variant and gene-based from multiple cohorts. However, it remains challenging to meta-analyze haplotype association tests from different cohorts with different haplotype structures. Here we propose a two-stage approach which allows cohorts to contribute association results from uniquely observed haplotypes, in addition to those observed in multiple cohorts. Association test of either single or overall haplotype effect/association can be obtained within our framework. A simulation study shows that the proposed approach has the correct type I error. We also present power evaluation for different scenarios. Our approach is applied to real exome-chip data. Hereby we assess the association between haplotypes formed by 16 variants in the known fasting glucose associated loci (G6PC2). Global association is detected (P=1.061674e-17) which is more significant than any single-variant analysis.