Case-Parent Triad Studies of Genetic Association and Gene-Gene Interaction in the Presence of Missing Data

• Activity Number: 235
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Tracy L. Bergemann*+ and Matthew Deyo-Svendsen
• Companies: University of Minnesota and University of Minnesota
• Keywords: Case-parent triad design ; Genetic interaction ; Genetic association ; Missing data

Abstract

Triad genotypes are analyzed with a variety of available methods, but we use log-linear models because they test for genetic association and estimate the relative risks of transmission. Our research addresses two concerns, (1) accounting for missing data without inflating type I error and (2) correcting bias when triad genotype combinations have zero counts. We perform a Multinomial-Poisson transformation and then optimize a trimmed likelihood. Maximum likelihood estimates give relative risks and their information matrix. A likelihood ratio test determines significance. Simulations demonstrate that these new methods properly maintain the type I error in a wide variety of scenarios. The trimmed likelihood successfully reduces the bias in relative risk estimates when the minor allele frequency is small. These result are especially important for the determination of interaction terms.