Biostatistics/Epidemiology/Research Design Core, Center for Clinical and Translational Sciences, The University of Texas Health Science Center at Houston, Houston, Texas 77030, USA.
Genet Epidemiol. 2011 Nov;35(7):658-63. doi: 10.1002/gepi.20615. Epub 2011 Aug 26.
In case-control genetic association studies, the robust procedure, Pearson's Chi-square test, is commonly used for testing association between disease status and genetic markers. However, this test does not take the possible trend of relative risks, which are due to genotype, into account. On the contrary, although Cochran-Armitage trend test with optimal scores is more powerful; it is usually difficult to assign the correct scores in advance since the true genetic model is rarely known in practice. If the unknown underlying genetic models are misspecified, the trend test may lose power dramatically. Therefore, it is desirable to find a powerful yet robust statistical test for genome-wide association studies. In this paper, we propose a new test based on the partition of Pearson's Chi-square test statistic. The new test utilizes the information of the monotonic (increasing or decreasing) trend of relative risks and therefore in general is more powerful than the Chi-square test; furthermore, it reserves the robustness. Using simulated and real single nucleotide polymorphism data, we compare the performance of the proposed test with existing methods.
在病例对照遗传关联研究中，皮尔逊卡方检验是一种常用的检验疾病状态与遗传标记之间关联的稳健方法。然而，该检验并未考虑到由于基因型导致的相对风险的可能趋势。相反，虽然最优评分的 Cochran-Armitage 趋势检验更具威力；但由于实际中很少知道真实的遗传模型，因此通常难以预先分配正确的评分。如果未知的潜在遗传模型被错误指定，那么趋势检验可能会显著失去效力。因此，对于全基因组关联研究，需要找到一种强大而稳健的统计检验方法。本文提出了一种基于 Pearson 卡方检验统计量划分的新检验方法。该新检验利用了相对风险单调（递增或递减）趋势的信息，因此通常比卡方检验更具威力；此外，它保留了稳健性。通过模拟和真实的单核苷酸多态性数据，我们将所提出的检验方法与现有方法的性能进行了比较。
