|
Abstract:
|
Genetic association studies can be broken into two basic categories: population-based studies in which unrelated subjects are recruited and family-based studies which rely on related subjects. The distinguishing characteristics of these two fundamental approaches have been addressed in a number of papers (Laird and Lange, 2006 and Hopper et al., 2005 among others). Recent technological and computational advances have made genome-wide association studies a reality (Herbert et al, 2006; Klein et al., 2005; Cheung et al., 2005) and have also made it necessary to consider the implications of employing different designs. Although these genome-wide studies are now feasible from a cost perspective, it is of utmost importance that they be designed to be cost-efficient, powerful and robust to modeling assumptions. We investigate here differences in statistical power between popular approaches.
|
