Quantile analysis on human complex traits in GWASand Sequencing studies.

描述（由申请人提供）：了解人类数量性状的遗传基础对公共卫生非常重要。主要的策略是识别影响感兴趣的数量性状的平均值的遗传变异。然而，高风险群体通常被确定为具有高或低值的数量性状的受试者。因此，研究与复杂性状上、下分位数的遗传关联更有意义。此外，最近的研究表明，遗传变异可以影响复杂性状的整个分布，并且它们的影响在不同的分位数上可能不同。因此，我们建议将分位数回归方法应用于GWAS和测序研究中的次要复杂性状。由于大多数GWAS和测序研究使用病例对照样本方案，因此它们不是一般人群的代表性样本。一般来说，简单估计的回归分位数可能会严重偏离真实的关联。从病例对照样本中恢复群体关联的统计方法被称为“二次分析”.这些方法大多数是基于似然的，并且仅估计性状的平均值的遗传效应。它们不能直接用于获得分位数估计值。为了进行一致和有效的估计， 在条件分位数上，我们提出了一个新的估计方程族，并开发了所有必要的统计工具，用于推断、变量选择和排序。我们将把所发展的方法应用于GWAS和测序研究，以调查与人类数量性状的遗传关联。所提出的工作有很大的潜力，深化和扩大现有的知识的遗传基础上的数量性状。

DESCRIPTION (provided by applicant): Understanding the genetic basis of human quantitative traits is of great importance for public health. The main strategy has been to identif genetic variants that influence the mean of a quantitative trait of interest. However, high-risk groups are often identified as the subjects who have either high or low values for their quantitative traits. Therefore, it is more meaningful to investigate the genetic association with the upper or lower quantiles of the complex traits. Moreover, recent studies indicate that genetic variants could influence the entire distributions of the complex traits, and their impact could differ at various quantiles. Hence, we propose to apply quantile regression methods to the secondary complex traits in GWAS and Sequencing studies. Since most GWAS and Sequencing studies use case-control sample schemes, they are not representative samples to the general population. Naively estimated regression quantiles could be substantially biased from the true association in general. Statistical methods recovering the population associations from case- control sample are known as "secondary analysis". Most of these methods are likelihood based, and only estimate the genetic effect on the means of the traits. They cannot be applied directly to obtain quantile estimates. In order to make consistent and efficient estimation on conditional quantiles, herein we propose a novel family of estimating equations, and also develop all the necessary statistical tools for inference, variable selection and ranking. We will apply the developed methods to GWAS and Sequencing studies to investigate the genetic association with human quantitive traits. The proposed work has great potential to deepen and expand the existing knowledge on the genetic basis of quantitative traits.