Statistically, this problem can be regarded as ‘‘weak power,’’ for which the primary solution is to increase the number of individuals in the study. This is not always technically or financially feasible. However, more and more DNA samples from patients with various diseases, as well as population control cohorts, are being genotyped and subsequently made publicly available through initiatives, such as the WTCCC, the Genetic Association Information Network (GAIN), and the database of genetic and phenotypic information (dbGaP). Thus, there is an opportunity to increase the statistical power of a study at no extra cost, by using the genotype data from these external samples to expand the primary “within-study” control cohort. Figure 1 presents the power of a GWA study of 500 cases to detect association of a causal SNP with risk allele frequency of 20% for a disease of prevalence 0.1%, with the number of control samples ranging from 500 to 5,000 individuals at a significance level of 5%. Power is presented as a function of the heterozygous genotype relative risk under the assumption of a multiplicative disease model.
