Zhu Degang, Hu Yue-Qing, Lin Shili
Department of Applied Mathematics, Nanjing Forestry University, Nanjing, China.
School of Statistics and Management, Shanghai University of Finance and Economics, Shanghai, China.
J Hum Genet. 2016 Nov;61(11):965-975. doi: 10.1038/jhg.2016.90. Epub 2016 Jul 14.
Although genome-wide association studies have successfully detected numerous associations between common variants and complex diseases, these variants typically can only explain a small part of the heritable component of a disease. With the advent of next-generation sequencing, attention has turned to rare variants. Recently, a variety of approaches for detecting associations of rare variants have been proposed, including the Kullback-Leibler divergence-based tests (KLTs) for detecting genotypic differences between cases and controls. However, few of these approaches consider linkage disequilibrium (LD) structure among rare variants and common variants. In this study, we propose two block-based association tests for testing the effects of rare variants on a disease. The main idea for this approach comes from the hypothesis that a region of interest may consist of two or more LD blocks such that single-nucleotide variants (SNVs) within each block are correlated, whereas SNVs in different blocks are independent or weakly correlated. Under this hypothesis, we propose two tests that are generalizations of the KLTs by taking the block structure into account. A simulation study under various scenarios shows that the proposed methods have well-controlled type I error rates and outperform some leading methods in the literature. Moreover, application to the Dallas Heart Study data demonstrates the feasibility and performance of the two proposed methods in a realistic setting.
尽管全基因组关联研究已成功检测出众多常见变异与复杂疾病之间的关联，但这些变异通常只能解释疾病遗传成分的一小部分。随着下一代测序技术的出现，人们的注意力已转向罕见变异。最近，已提出了多种检测罕见变异关联的方法，包括基于库尔贝克-莱布勒散度的检验（KLTs），用于检测病例组和对照组之间的基因型差异。然而，这些方法中很少有考虑罕见变异和常见变异之间的连锁不平衡（LD）结构的。在本研究中，我们提出了两种基于模块的关联检验，用于检验罕见变异对疾病的影响。这种方法的主要思想源于这样一个假设：感兴趣的区域可能由两个或更多的LD模块组成，使得每个模块内的单核苷酸变异（SNV）是相关的，而不同模块中的SNV是独立的或弱相关的。在这个假设下，我们提出了两种检验方法，它们是通过考虑模块结构对KLTs的推广。在各种情况下进行的模拟研究表明，所提出的方法具有良好控制的I型错误率，并且优于文献中的一些领先方法。此外，应用于达拉斯心脏研究数据证明了所提出的两种方法在实际环境中的可行性和性能。
