Methods for leveraging family-based designs and summary data to elucidate complex trait genetics

摘要 为了更好地理解复杂的人类特征的遗传基础，这两个根本不同和互补的特征 全基因组关联研究中采用的设计是基于群体和基于家庭的设计。 随着生物库和大规模生物医学数据库的出现，近年来 对成人特征/疾病的遗传研究，因此，人口方法学的快速进步-- 这些生物库所依赖的基础设计。相比之下，基于家庭的设计方法几乎没有收到什么效果 尽管它们在低患病率/罕见的遗传基础研究中发挥了重要作用，但仍未引起注意 儿童疾病和儿童健康后果的影响。基于家庭设计的分析方法可以防止 人口分层和混杂(从而允许参与者之间的种族/民族多样性)，并可以 比类似样本量的基于总体的研究更有说服力。大规模经济衰退的另一个后果 BIOBANKS是聚合水平的基因型-性状关联结果的公共可获得性(或GWAS摘要 统计学)广泛的复杂的人类特征，包括介于 基因和与疾病相关的特征。可以利用GWAS汇总统计数据来理解生物学的方法 潜在疾病的需求量很大，因为它们几乎同样有效，并避免了后勤/伦理方面的问题 与共享个人级别的数据相关。在这一应用中，我提出了一个开发的研究方案 遗传流行病学研究的新统计方法和开放获取工具，特别是 注重以家庭为基础的设计。其中一些方法/工具将仅利用关联摘要 统计学创新地将组学与疾病数据相结合，从而有助于提高对 作为人类健康基础的调控机制。我们试图解决人类的一些公开问题 性状遗传学，包括识别非加性遗传效应(例如基因-基因)的方法学挑战 相互作用、基因-环境相互作用、亲本起源效应)、稀有变异的影响以及优先顺序 通过整合组学的因果变异。我们将从统计文献中引入晦涩难懂的数学函数 到真正的公共卫生应用，同时在现有的数据库上进行说明。这项研究计划将 以三种不同的方式支持多样性：以家庭为基础的设计的方法进步，以克服 与参与者种族/民族多样性有关的挑战；允许基因组研究人员使用有效的方法/工具 即使在资源匮乏的地方，也要使用公开的摘要数据进行遗传流行病学研究 并帮助培养约翰·霍普金斯大学公共学院每年招收的不同类型的研究生 健康。在过去的5年里，我在以家庭为基础的遗传学研究中与人口研究一起建立了研究概况- 开发了基于汇总级数据的尖端方法，实现了数据驱动 通过可复制的数据科学方法/工具制定政策，并获得指导技能。我的多- 纪律训练和我以前的经验使我处于一个独特的位置，可以成功完成这个项目。

ABSTRACT To better understand genetic basis of complex human traits, two fundamentally different and complementary designs employed in genome-wide association studies (GWAS) are population-based and family-based designs. With the advent of biobanks and large-scale biomedical databases, recent years have seen an explosion in genetic studies of adult traits/diseases, and consequently, a rapid advancement in methodology for population- based designs that these biobanks depend on. In contrast, methods for family-based designs have received little to no attention although they play an important role in the investigation of genetic basis of low-prevalence/rare disorders and of child health outcomes. Analysis methods based on family-based designs can protect against population stratification and admixture (thus allowing for racial/ethnic diversity among participants), and can be more powerful than a population-based study of similar sample size. Another consequence of large-scale biobanks is the public availability of aggregate-level genotype-trait association results (or GWAS summary statistics) for a wide spectrum of complex human traits, including molecular traits that are intermediate between genotype and a disease-related trait. Methods that can leverage GWAS summary statistics to understand biology underlying diseases are in high demand since they are nearly as efficient and avoid logistical/ethical concerns related to sharing individual-level data. In this application, I propose a research program of developing novel statistical methods and open-access tools for genetic epidemiology studies, with a particular focus on family-based designs. Some of these methods/tools will leverage only association summary statistics to innovatively integrate omics with disease data, thereby helping improve understanding of regulatory mechanisms underlying human health. We seek to address some of the open problems of human trait genetics, including methodological challenges in identifying non-additive genetic effects (e.g. gene-gene interaction, gene-environment interaction, parent-of-origin effect), effects of rare variants, and in prioritizing causal variants through integrative omics. We will bring obscure mathematical functions from statistical literature to real public health applications while illustrating them on existing databases. This research program will support diversity in three distinct ways: methodological advancement of family-based designs that overcome challenges related to racial/ethnic diversity in participants; efficient methods/tools that allow genomic researchers to conduct genetic epidemiology studies using publicly available summary data even in resource-poor environments; and help train diverse graduate students recruited annually by the Johns Hopkins School of Public Health. In the last 5 years, I have built a research profile in family-based genetic studies alongside population- based ones, have developed cutting-edge methods based on summary-level data, have enabled data-driven policy-making via reproducible data science methods/tools, and have acquired mentoring skills. My multi- disciplinary training and my prior experience puts me in a unique position to successfully complete this program.