Department of Biostatistics, Harvard School of Public Health, Boston, MA 02115, USA.
BMC Genet. 2005 Dec 30;6 Suppl 1(Suppl 1):S115. doi: 10.1186/1471-2156-6-S1-S115.
Due to the recent gains in the availability of single-nucleotide polymorphism data, genome-wide association testing has become feasible. It is hoped that this additional data may confirm the presence of disease susceptibility loci, and identify new genetic determinants of disease. However, the problem of multiple comparisons threatens to diminish any potential gains from this newly available data. To circumvent the multiple comparisons issue, we utilize a recently developed screening technique using family-based association testing. This screening methodology allows for the identification of the most promising single-nucleotide polymorphisms for testing without biasing the nominal significance level of our test statistic. We compare the results of our screening technique across univariate and multivariate family-based association tests. From our analyses, we observe that the screening technique, applied to different settings, is fairly consistent in identifying optimal markers for testing. One of the identified markers, TSC0047225, was significantly associated with both the ttth1 (p = 0.004) and ttth1-ttth4 (p = 0.004) phenotype(s). We find that both univariate- and multivariate-based screening techniques are powerful tools for detecting an association.
由于单核苷酸多态性数据可用性的提高，全基因组关联测试成为可能。人们希望这些额外的数据可以确认疾病易感基因座的存在，并确定疾病的新遗传决定因素。然而，多重比较的问题可能会削弱这些新数据带来的任何潜在收益。为了规避多重比较问题，我们使用了一种最近开发的基于家系的关联测试筛选技术。这种筛选方法可以在不影响我们检验统计量的名义显著性水平的情况下，确定最有前途的单核苷酸多态性进行检验。我们比较了单变量和多变量基于家系的关联测试中筛选技术的结果。从我们的分析中，我们观察到，应用于不同环境的筛选技术在识别最佳测试标记方面相当一致。一个被识别的标记 TSC0047225 与 ttth1（p=0.004）和 ttth1-ttth4（p=0.004）表型均显著相关。我们发现，单变量和多变量的筛选技术都是检测关联的有力工具。
