|
Activity Number:
|
153
- Recent Developments in Functional Data Analysis and Empirical Likelihood Methods in Biostatistics
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Monday, July 31, 2017 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
|
Abstract #323921
|
View Presentation
|
|
Title:
|
Gene-Based Association Testing of Dichotomous Traits with Generalized Functional Linear Mixed Models Using Extended Pedigrees
|
|
Author(s):
|
Ruzong Fan* and Yingda Jiang and Chi-yang Chiu and Momiao Xiong and Christopher I Amos and Daniel Weeks and Richard John Cook and M'Hamed Lajimi Lakhal-Chaieb and Qi Yan and Wei Chen and Michael B. Gorin and Yvette P. Conley and Alexander F. Wilson and Joan Bailey-Wilson and McMahon Francis
|
|
Companies:
|
and University of Pittsburgh and NIH and University of Texas and Geisel School of Medicine at Dartmouth and University of Pittsburgh and University of Waterloo and Université Laval and Division of Pulmonary Medicine, Allergy and Immunology, Children's Hospital of Pittsburgh and University of Pittsburgh and UCLA and University of Pittsburgh and NIH and NIH and NIH
|
|
Keywords:
|
complex diseases ;
extended pedigree ;
generalized functional linear mixed models ;
rare variants ;
common variants ;
association study
|
|
Abstract:
|
In recent years, fixed effect functional regression models have been developed to test for gene-based association between a complex trait and multiple genetic variants. The functional regression models reduce high dimensionality of next-generation sequencing data to draw useful information. Here we extend this approach to accommodate family-based data using generalized functional linear mixed models (GFLMM) to test for association between a dichotomous trait and genetic variants in a defined chromosome region. Gene-based association tests are constructed by using the GFLMM to model the effect of a major gene as a fixed mean, the polygenic contributions as a random variation, and the correlation of pedigree members by kinship coefficients. Then likelihood ratio test (LRT) statistics are used to test if the fixed mean is zero. Simulation results indicate that the LRT GFLMM statistics accurately control the type I error rates. The LRT statistics have similar or higher power than the retrospective kernel and burden statistics developed by Schaid and colleagues. We illustrate the behavior of these statistics by applying them to an age-related macular degeneration family dataset.
|
Authors who are presenting talks have a * after their name.
