Two-stage Case-control Designs In Genome-wide Association Study

In recent years, genome-wide association studies (GWAS) are being conducted to unravel the genetic etiology of complex human diseases. There are three main issues in GWAS design: power of test, false positive rate and cost. Since hundreds of thousands single nucleotide polymorphisms(SNPs) would be genotyped for each individual or sample, one-stage case-control designs may not be cost-effective. In practice there are situations which the available budget is not enough for genome-wide genotyping all of the sample. Therefore, an optimial design of such studies to maximize the power to identify loci of true association was needed. In this thesis, the author proposed an alternative two-stage case-control designs in GWAS.In this thesis, the author introduced the principles of two-stage case-control design, and its implementing steps in genome-wide association study. The simulation studies were conducted to evaluate the statistical power for detecting the effects of SNPs, family-wise type I error rate (FWER) and false discovery rate (FDR) of one-stage and two-stage case-control studies in GWAS when the total cost of genotyping remain constant. The minor allele frequency (MAF), genotype relative risk, the ratio of stage 2 to stage 1 per genotype cost were also considered in the simulation studies.The main conclusions of this study are as following:(1) Two-stage case-control studies are more powerful than one-stage case-control studies in GWAS when the total cost of genotyping remain constant.(2) The power of two-stage case-control study would decrease if the cost of second stage is increases.(3) Both of one-stage and two-stage designs′FWER are around the 0.05, which is same as pre-specified significance level, while the FDR of two-stage designs is lower than the corresponding one of one-stage designs.