Cheng Kuang-Fu, Lee Jen-Yu
Biostatistics Center and School of Public Health, Taipei Medical University, Taipei, Taiwan, ROC.
Hum Hered. 2014;78(1):38-46. doi: 10.1159/000360987. Epub 2014 Jun 21.
There are several well-known single SNP tests presented in the literature for detecting gene-disease association signals. Having in place an efficient and robust testing process across all genetic models would allow a more comprehensive approach to analysis. Although some studies have shown that it is possible to construct such a test when the variants are common and the genetic model satisfies certain conditions, the model conditions are too restrictive and in general difficult to verify. In this paper, we propose a powerful and robust test without assuming any model restrictions. Our test is based on the selected 2 × 2 tables derived from the usual 2 × 3 table. By signals from these tables, we show through simulations across a wide range of allele frequencies and genetic models that this approach may produce a test which is almost uniformly most powerful in the analysis of low- and high-frequency variants. Two cancer studies are used to demonstrate applications of the proposed test.
文献中提出了几种用于检测基因与疾病关联信号的著名单核苷酸多态性（SNP）测试。在所有遗传模型中建立一个高效且稳健的测试过程，将允许采用更全面的分析方法。尽管一些研究表明，当变异体常见且遗传模型满足某些条件时，构建这样一个测试是可能的，但模型条件过于严格，一般难以验证。在本文中，我们提出了一种强大且稳健的测试方法，无需假设任何模型限制。我们的测试基于从通常的2×3表中导出的选定2×2表。通过这些表中的信号，我们通过在广泛的等位基因频率和遗传模型上进行模拟表明，这种方法可能产生一种在分析低频和高频变异体时几乎具有一致最强功效的测试。两项癌症研究被用于证明所提出测试的应用。
