Integrative multivariate association and genomic analyses

描述（由申请人提供）：为了了解人类复杂疾病的基因组结构和病因，在过去的几十年里，人们在全基因组遗传变异、转录组和其他基因组信息的研究上付出了巨大的努力。迄今为止，已经产生了丰富的资源，其中大多数是在针对各自的主要目标/假设进行分析后公开提供的。然而，我们对人类疾病机制的理解才刚刚开始，这些理解将需要识别遗传和表观遗传风险因素的骨干，并将关键因素整合到一个协同系统中。为了最好地利用现有数据并促进对复杂人类疾病的研究，拟议研究的长期目标是发展强大和有效的统计方法和计算工具，主要在两个领域进行多变量分析：通过整合基因组和非基因组信息进行关联研究，以便进一步确定复杂疾病的遗传变异；整合基因组分析，联合分析基因组中的遗传变异、转录组和其他信息。在目标1中，我们提出了基于基因的关联测试的新颖而强大的方法，用于识别与多变量疾病谱相关的遗传变异，以及基于基因的基因-环境相互作用测试。在目标2中，我们开发了用于构建和比较eQTL网络的正则化方法。后者还可以通过表征不同表型或环境组的遗传调控模式变化来揭示重要的遗传变异和调控关系。我们提出的大部分工作是由砷毒性的遗传-基因组研究，基因-环境多表型研究（GEMS）所激发的，并将应用于该研究。在目标3中，我们提出了针对该数据集特征量身定制的方法；我们还将在这项独特的大型砷毒性研究中测试新的科学假设。我们的建议具有成本效益，因为它分析了GEMS的现有数据，同时为新的研究方向提供了方法和工具。我们预计，拟议的方法开发，当应用于砷毒性数据时，将对临床试验治疗效果和几种复杂疾病/特征的疾病病因学产生有价值的见解，包括但不限于砷相关的皮肤癌，心血管疾病激素测量，体重指数和血压。

DESCRIPTION (provided by applicant): In order to understand the genomic architecture and etiology for complex human diseases, great efforts have been extended in the past decades on research involving genome-wide genetic variation, transcriptome, and other genomic information. To date, rich resources have been generated and most are made publicly available after being analyzed for respective primary goals/hypotheses. Yet our understandings of human disease mechanisms are just beginning, and those understandings would require both the identification of a cadre of genetic and epigenetic risk factors, and the integration of key factor into a synergistic system. To best utilize existing data and facilitate research on complex human diseases, the long-term objective of the proposed research is to develop powerful and efficient statistical methods and computational tools for multivariate analyses in mainly two areas: association studies with the integration of genomic and non-genomic information in order to further identify genetic variation for complex diseases; and integrative genomic analyses that jointly analyze genetic variation, transcriptome, and other information in the genome. In Aim 1, we propose novel and powerful methods for gene-based association tests, for identification of genetic variation associated with multivariate disease profiles, and for gene-based gene-environment interaction tests. In Aim 2, we develop regularized methods for construction and comparison of eQTL networks. The later can also be used to reveal important genetic variants and regulatory relationships through characterizing the changes in genetic regulatory patterns across different phenotypic or environmental groups. Much of our proposed work is motivated by and will be applied to a genetic-genomic study on arsenic toxicity, Gene-Environment Multi- phenotype Study (GEMS). In Aim 3, we propose methods tailored for the characteristics of this data set; we will also test novel scientific hypotheses on this unique and large arsenic toxicity study. Our proposal is cost- effective as it analyzes existing data from GEMS while providing methods and tools for new research directions. We anticipate that the proposed method development, when applied to and beyond the arsenic toxicity data, would yield valuable insights on clinical trial treatment effects, and on disease etiology for several complex diseases/traits, including but not limited to, arsenic-related skin cancer, cardiovascular diseases hormone measures, body mass index and blood pressure.