Power Calculations for Genetic Association Studies Using Estimated Probability Distributions

Abstract

The determination of the power of-or of an appropriate sample size for-genetic association studies that exploit linkage disequilibrium requires many assumptions. Some of the more important assumptions include the linkage-disequilibrium strength among alleles at the observed marker-locus sites and a potential trait-influencing locus, the frequencies of the marker locus and trait-influencing alleles, and the ultimate density of the marker locus "map" (i.e., the number of bases between marker loci) necessary in order to identify, with some confidence, trait-influencing alleles. I consider an approach to assessment of the power and sample-size requirements of genetic case-control association study designs that makes use of empirically derived estimates of the distributions of important parameters often assumed to take on arbitrary values. My proposed methodology is extremely general and flexible and ultimately can provide realistic answers to questions such as "How many markers and/or how many individuals might it take to identify, with confidence, a disease gene, via linkage-disequilibrium and association methods from a candidate region or whole genome perspective?" I showcase aspects of the proposed methodology, using information abstracted from the literature.