Supplementary Material for: Statistical Models for Haplotype Sharing in Case-Parent Trio Data

Background: Haplotype sharing statistics have been introduced in an ad-hoc way, often relying heavily on permutation testing. As a result, applying these approaches to whole genome association studies or to evaluate their properties in extensive simulation experiments is problematic. Further, permutation testing may be inappropriate in the presence of phase ambiguity and population stratification. Aims: To present a simple framework for a class of haplotype sharing statistics useful for association mapping in case-parent trio data. This framework allows derivation of novel haplotype sharing tests as well as simple variance estimators and asymptotic distributions for haplotype sharing tests. Results and Conclusions: We validated that our approach is appropriately sized using simulated data, and illustrate the methodology by analyzing a Crohn��s disease dataset. We find that haplotype-based analyses are much more powerful than single-locus analyses for these data.背景：单倍型共享统计量（haplotype sharing statistics）的提出多为特设方式，且往往过度依赖置换检验（permutation testing）。这导致此类方法难以应用于全基因组关联研究（whole genome association studies），也无法在大规模模拟实验中对其性能进行有效评估。此外，当存在相位歧义（phase ambiguity）与人群分层（population stratification）时，置换检验的适用性并不恰当。研究目标：针对病例-父母三联体数据（case-parent trio data）中的关联定位（association mapping）任务，提出一类适用于单倍型共享统计量的简易分析框架。该框架可推导出新型单倍型共享检验方法，同时为单倍型共享检验提供简洁的方差估计量（variance estimators）与渐近分布（asymptotic distributions）。结果与结论：本研究通过模拟数据验证了所提方法的检验效能合理性，并通过分析克罗恩病（Crohn’s disease）数据集对该方法进行了实例演示。研究结果表明，针对该数据集，基于单倍型的分析方法的效能远高于单基因座分析（single-locus analyses）。