Statistical Methods in Genetic Association Studies

Abstract

Population structure is a serious confounding factor in genetic association studies. It may lead to false positive results or failure to detect true association. We propose a hierarchical clustering algorithm, AW-clust, for using single nucleotide polymorphism (SNP) genetic data to assign individuals to populations. We show that the algorithm can assign sample individuals highly accurately to their corresponding ethic groups: CEU, YRI, CHB+JPT in our tests using HapMap SNP data and it is also robust to admixed populations when tested on Perlegen SNP data. Moreover, it can detect fine-scale population structure as subtle as that between Chinese and Japanese by using genome-wide hight diversity SNP loci. Genotyping errors exist in most genetic data and can influence the biological conclusions of the studies. A simple method is to conduct the Hardy-Weinberg equilibrium (HWE) test in population-based association studies. We investigated the power issue of using the HWE test on genotyping error detection in the presence of current genotyping technologies. Multiple testing is a challenging issue in genetic studies using SNPs that are in linkage disequilibrium (LD) with each other. Failure to adjust for multiple testing appropriately may produce excess false positives or overlook true positive signals. We propose a new multiple testing correction method, CLDMeff , for association studies using SNP markers. It is shown to be simpler and more accurate than the recently developed methods and is comparable to the permutation-based correction using both simulated and real data. The efficiency and accuracy of the CLDMeff method makes it an attractive choice for multiple testing correction when there is high intermarker LD in the SNP dataset.