Family-based rare variant association methods for quantitative traits

描述(申请人提供)：高通量测序平台的技术进步使得通过全基因组测序和定向外显子测序将全基因组关联研究(GWAS)扩展到稀有变种成为可能。定制芯片，如免疫芯片和代谢芯片，因其低成本而被用于对候选感兴趣区域的稀有变异进行基因分型。目前，大多数复杂性状的测序项目都集中在无关的个体上。尽管基于家系的设计在稀有变异关联分析中起着重要作用，但实际上还没有开发出通用的统计方法来分析家系中复杂性状的稀有变异数据。我们建议开发强大而稳健的统计方法来测试感兴趣的基因组区域中的多个稀有变异的联合效应与家系数据中的数量性状之间的关联。将最近发展的序列核关联检验(SKAT)扩展到家族数据。我们提出的稀有变异关联方法在存在家族结构和/或人口分层的情况下将具有适当的I型错误率，并且将对感兴趣的基因组区域中稀有变异的大小和影响方向的潜在异质性具有健壮性。关联的统计意义将被分析评估，从而绕过在存在家族相关性的情况下设计适当的排列过程的困难。我们计划通过大规模的模拟研究来检验我们提出的方法在各种现实场景下的性能。我们的方法将在免费分发的软件中实现，允许其他研究人员直接应用这些方法来分析他们自己的数量性状罕见变异数据。

DESCRIPTION (provided by applicant): Technological advances in high-throughput sequencing platforms have made it possible to extend genome-wide association studies (GWAS) to rare variants by whole-genome sequencing and targeted exome-sequencing. Custom chips such as the ImmunoChip and MetaboChip have been utilized for their low cost in genotyping rare variants in candidate regions of interest. Currently most sequencing projects for complex traits are focused on unrelated individuals. Despite the important role the family-based design plays in the rare variant association analysis, there are virtually no general statistical methods developed to analyze rare variant data for complex traits in families. We propose to develop powerful and robust statistical methods to test for association between the joint effects of multiple rare variants in a genomic region of interest and a quantitative trait in family data. e extend the recently developed sequence kernel association test (SKAT) to family data. Our proposed rare variant association methods will have appropriate type I error rates in the presence of family structure and/or population stratification, and will be robust to potential heterogeneity in size and directions of effect in rare variants across a genomic region of interest The statistical significance of association will be assessed analytically, circumventing the difficulty of designing an appropriate permutation procedure in the presence of familial correlation. We plan to examine performance of our proposed methods through large-scale simulation studies under a wide range of realistic scenarios. Our methods will be implemented in freely distributed software, allowing other investigators to apply the methods directly to analysis of their own rare variant data for quantitative traits.