Association Tests of Rare Variants Using Sequence Reads without Calling Genotypes

 描述（由申请人提供）：下一代DNA测序（NGS）为发现影响罕见疾病的变异提供了前所未有的机会。然而，当前的做法是首先调用潜在的基因型，然后将调用的值视为罕见变异体测试中已知的值，这在存在基因分型错误的情况下是有问题的，并且如果基因分型差的变异体被完全过滤掉，则效率低下。本申请的目标是开发超越标准基因型调用范式的统计方法，以替代直接对测序读数进行建模，用于测试罕见变异关联。我们相信我们的方法对NGS研究的许多实际设计是稳健的，并且比仅使用其基因型被准确调用的变体更强大。目的1提出了在具有外部对照的病例对照研究中测试罕见变异关联的方法，我们相信这对病例和对照之间的系统性测序差异是稳健的。目的2提出了罕见的变异的方法，为病例-父母三人组研究，控制I型错误，制定协会通过潜在的传输罕见的等位基因。目的3建立了新的方法来测试罕见的变异关联的X染色体;我们的可能性方法的灵活性，使其成为理想的解释性别之间的拷贝数差异，X失活，不同的性别比例之间的情况下，控制，和性别特异性的影响。我们将使用来自我们积极参与的真实的测序研究的数据集来评估这些方法，并将在用户友好的软件中实现这些方法以供公开分发（目标4）。

 DESCRIPTION (provided by applicant): Next-generation sequencing (NGS) of DNA provides an unprecedented opportunity to discover rare disease- influencing variants. However, the current practice of first calling the underlying genotypes and then treating the called values as known in rare variant tests is problematic in the presence of genotyping errors and inefficient if poorly genotyped variants are filtered out altogether. The goal of this application is to develop statistical approaches that move beyond the standard genotype-calling paradigm to instead model the sequencing reads directly for testing rare variant associations. We believe our approaches are robust to many practical designs of NGS studies, and substantially more powerful than the use of only variants whose genotypes are accurately called. Aim 1 proposes methods for testing rare variant associations in case-control studies with external controls, which we believe will be robust to the systematic sequencing differences between cases and controls. Aim 2 proposes rare variant methods for case-parent trio studies that control the type I error by formulating association through the underlying transmissions of the rare allele. Aim 3 establishes novel methods for testing rare variant associations on the X chromosome; the flexible nature of our likelihood approach makes it ideal for accounting for copy number difference between sexes, X inactivation, different sex ratios between cases and controls, and sex-specific effects. We will evaluate these methods using datasets from real sequencing studies that we are actively involved in and will implement the methods in user- friendly software for public distribution (Aim 4).