A test of Hardy-Weinberg equilibrium in structured populations

Testing for Hardy-Weinberg equilibrium (HWE) is used routinely as an important initial step for genotype data quality checking. Departure from HWE can be caused by many factors, such as genotyping errors, population stratification, and disease association, if we use affected individuals only. In a structured population, even if a marker is in HWE in each subpopulation, data may show departure from HWE if allele frequencies are different in different subpopulations and such a departure can be misinterpreted as a potential problem in genotyping quality, resulting in false exclusion from future analysis. In this article, we propose a new HWE test, a test for HWE in structured populations (HWES) that can assess departure from HWE and take into account of population stratification at the same time. Our proposed test can distinguish departure from HWE caused by population stratification and departure from HWE caused by other factors. We use simulation studies as well as applications to real data sets to evaluate the performance of the proposed test. Results show that, for a wide range of population structures, our proposed test has correct type I error rates while the traditional χ(2) test will lead to false-positive results. In homogenous populations, our proposed test has comparable power with the traditional χ(2) test.