Lange Kenneth, Sinsheimer Janet S, Sobel Eric
Department of Biomathematics, David Geffen School of Medicine at UCLA, Los Angeles, California 90095-7088, USA.
Genet Epidemiol. 2005 Jul;29(1):36-50. doi: 10.1002/gepi.20073.
This report presents an overview of association testing strategies from a user's perspective, with particular attention to the capabilities of the computer program Mendel. Association testing is driven by the nature of the study sample, the nature of the disease trait, and the kind of markers employed. The practicing statistician must also choose whether to conduct parametric or nonparametric tests. Because of the complexities involved, Mendel offers users several analysis options. The different options are tied together by shared input and output conventions and a shared language for defining models. Mendel also features new statistics and theory found in no other genetics software. The most important innovations include: association testing by penetrance estimation, expansion of matched-pair designs to permutation unit designs, and a rigorous implementation of the measured genotype approach for quantitative trait loci. This report explains how Mendel imputes allele counts and conducts both asymptotic and permutation tests in the measured genotype framework.
本报告从用户角度概述了关联测试策略，特别关注计算机程序Mendel的功能。关联测试受研究样本的性质、疾病性状的性质以及所采用标记的类型驱动。执业统计学家还必须选择进行参数检验还是非参数检验。由于涉及的复杂性，Mendel为用户提供了几种分析选项。不同的选项通过共享的输入和输出约定以及用于定义模型的共享语言联系在一起。Mendel还具有其他遗传学软件中未发现的新统计数据和理论。最重要的创新包括：通过外显率估计进行关联测试、将配对设计扩展到置换单元设计，以及对数量性状位点严格实施测量基因型方法。本报告解释了Mendel如何估算等位基因计数，并在测量基因型框架中进行渐近检验和置换检验。
