Purcell Shaun, Sham Pak, Daly Mark J
Whitehead Institute for Biomedical Research, Cambridge, MA, USA.
Am J Hum Genet. 2005 Feb;76(2):249-59. doi: 10.1086/427886. Epub 2004 Dec 21.
Family-based association designs are popular, because they offer inherent control of population stratification based on age, sex, ethnicity, and environmental exposure. However, the efficiency of these designs is hampered by current analytic strategies that consider only offspring phenotypes. Here, we describe the incorporation of parental phenotypes and, specifically, the inclusion of parental genotype-phenotype correlation terms in association tests, providing a series of tests that effectively span an efficiency-robustness spectrum. The model is based on the between-within-sibship association model presented in 1999 by Fulker and colleagues for quantitative traits and extended here to nuclear families. By use of a liability-threshold-model approach, standard dichotomous and/or qualitative disease phenotypes can be analyzed (and can include appropriate corrections for phenotypically ascertained samples), which allows for the application of this model to analysis of the commonly used affected-proband trio design. We show that the incorporation of parental phenotypes can considerably increase power, as compared with the standard transmission/disequilibrium test and equivalent quantitative tests, while providing both significant protection against stratification and a means of evaluating the contribution of stratification to positive results. This methodology enables the extraction of more information from existing family-based collections that are currently being genotyped and analyzed by use of standard approaches.
基于家系的关联设计很受欢迎，因为它们能对基于年龄、性别、种族和环境暴露的人群分层进行内在控制。然而，目前仅考虑后代表型的分析策略阻碍了这些设计的效率。在此，我们描述了纳入亲代表型，特别是在关联测试中纳入亲代基因型-表型相关项，提供了一系列能有效跨越效率-稳健性范围的测试。该模型基于1999年富尔克及其同事提出的用于数量性状的同胞间关联模型，并在此扩展到核心家庭。通过使用阈值模型方法，可以分析标准的二分法和/或定性疾病表型（并且可以包括对表型确定样本的适当校正），这使得该模型能够应用于常用的患病先证者三人组设计的分析。我们表明，与标准传递/不平衡测试和等效的定量测试相比，纳入亲代表型可以显著提高检验效能，同时提供对分层的显著保护以及评估分层对阳性结果贡献的方法。这种方法能够从目前正在使用标准方法进行基因分型和分析的现有基于家系的样本集中提取更多信息。
