Statistics and Its Interface
Volume 7 (2014)
We dedicate this special issue to Dr. Gang Zheng, a great colleague and dear friend to many of us.
A robust test for quantitative trait analysis with model uncertainty in genetic association studies
Pages: 61 – 68
Statistical tests that assume an additive model are commonly employed in genetic association studies. However, the true models for genetic variants are rarely known. A mis-specified genetic model may lead to loss of power in identifying the potential markers associated with a disease. In this paper, we develop a robust test based on modified $F$-test statistics for quantitative trait genetic association studies and a simple method to compute its statistical significance and power. We also study sample size calculations for designing such an association study. Numerical results, including simulation studies and a real data example, show that the proposed robust test has satisfactory performance when the model is unknown and is more robust than some existing procedures when the model is mis-specified.
$F$-test, robust, quantitative trait, genome-wide association studies