Developing Stats Methods to Detect Rare Genetics Variants in Human Pedigrees

新一代测序技术结合高效的DNA捕获方法为研究复杂表型的遗传基础提供了外显子组测序方法。与全基因组关联研究（GWAS）不同，它只能发现人群中常见的DNA变异（大于1%），外显子组测序是今天可能有兴趣寻找罕见突变的科学家的绝佳选择。此外，外显子组测序具有全面测试常见和罕见的编码变异的作用的优势。预计每个基因都可能含有功能相关的变体。 最近，一些统计方法成为可用于分析的贡献，罕见的变异复杂性状的发展。这些方法包括多变量合并塌陷法（CMC）、多变量塌陷亚组检验、Hotelling T2检验、MANOVA、Fisher乘积法、加权求和法和基于核的自适应检验。 虽然这些方法的优点已被广泛评估为基于人群的关联研究，这些方法在其目前的形式可以被用来分析基于谱系的关联分析使用外显子组测序数据。 我们将开发基于家系的罕见变异分析方法，将每个受影响的亲属视为依赖对，并使用相关矩阵来解释依赖关系。 在一组罕见变异与疾病无关的零假设下，我们开发的新统计量是渐近分布的中心分布。 此外，我们将使用估计的基于IBD的权重来解释从相同谱系生成的相关受影响或未受影响对的依赖性。该方法将用于分析具有外显子组数据的双相情感障碍谱系。模拟研究将用于确定功效和I类误差。该方法将用于分析约100个双相情感障碍家系的外显子组数据，以及未来与其他精神疾病的数据集。

Next-generation sequencing technologies coupled with the efficient DNA capture methods provide exome sequencing approach to investigate the genetic basis of complex phenotypes. Unlike whole genome association studies (GWAS) which can only discover variation in DNA that is frequent in the population (great than 1%), exome sequencing is a great choice for scientists today who might be interested in looking for rare mutations. Furthermore, exome sequencing has the advantage of testing comprehensively the role of coding variation, both common and rare. It is anticipated that every gene may harbor functionally relevant variants. Recently, a number of statistical methods become available for analyzing the contribution of rare variants to the development of complex traits. These methods include Combined Multivariate and Collapsing (CMC) Method, Multivariate test of collapsed sub-groups Hotelling T2 test, MANOVA, Fishers product method, Weighted Sum Method and Kernel-based adaptive test. While the merits of these methods have been evaluated extensively for population-based association studies, none of these methods in their current form can be used to analyze the pedigree based association analysis using exome sequencing data. We will develop pedigree-based rare variants analysis approach by treating each affected relatives as dependent pairs and the dependency will be accounted for using correlation matrix. Under the null hypothesis of no association of a set of rare variants with the diseases, the new statistic we have developed is asymptotically distributed as a central distribution. Further, we will use the estimated IBD based weights to account for the dependency of the related affected or unaffected pairs generated from same pedigrees. This method will be used to analyze approximately bipolar disorder pedigrees with exome data. Simulation studies will be used for determining power and type I errors. This method will be used to analyze approximately 100 bipolar disorder pedigrees with exome data as well as data sets with other mental disorders in the future.