Reducing sample sizes in genome scans: Group sequential study designs with futility stops

Optimum two-stage designs in case-control association studies using false discovery rate

Genetic association studies using case-control designs are often done to identify loci associated with disease susceptibility. These studies are often expensive to perform, due to a large number of genetic markers. Several types of two-stage designs are proposed and used from the point of cost effectiveness. We proposed to control the false discovery rate for multiple-testing correction in two-stage designs, using optimal sample sizes and criteria for selecting markers associated with a disease in each stage to minimize the cost of genotyping. The expected power and cost of two-stage designs were compared with those of one-stage designs, under the assumptions that the genetic markers are independent and total sample size is fixed. The results showed that the proposed two-stage procedure usually reduced the cost of genotyping by 40-60%, with a power similar to that of the one-stage designs. In addition, the sample size and selection criteria, which are optimized parameters, are defined as a function of a prior probability that marker-disease association is true. So, the effects of mis-specification of a prior probability on efficiency were also considered.