Mapping Multiple Complex and Omics Trait-Associations Using Summary Statistics

项目摘要 对于许多与疾病相关的遗传变异，变异影响的作用机制 疾病易感性尚不清楚。因为遗传变异也会影响分子表型，如 转录组、甲基组和蛋白质组，研究“组学”结果可能有助于更好地理解 疾病的进程。特别是，对多组学数据的联合分析可能会增强我们对 基因对这些结果的影响是在一个多水平的分子系统中协调的，从而导致疾病 敏感度。因为对分子表型的遗传效应可能进一步取决于组织、细胞类型或其他 在这种情况下，科学界将受益于继续开发方法，将多个 不同条件或背景的组学数据。然而，庞大的数据规模加上未知 跨特征或条件的相关性结构使这样的分析具有挑战性。在这个项目中，我们 提出有效的方法来整合来自复杂遗传效应的多项研究的汇总统计数据 和组学表型。对现有的多组学综合方法进行总结改进 作为输入，我们将联合分析扩展到三种以上的数据类型或条件，并允许 统计数据来自重叠的样本。应用程序中提供的初步结果表明， 所提出的方法在计算上是可行的，并且产生的结果与当前的结果一致。 生物学知识。这些方法的拟议应用有可能识别新的关联或 为组学特征和癌症风险之间的已知关联提供新的证据。这件事的成功 工作将提供灵活的方法和计算工具，可以应用于其他疾病和环境。

Project Summary For many disease-associated genetic variants, the functional mechanism through which the variant affects disease susceptibility is unknown. Because genetic variants also affect molecular phenotypes such as the transcriptome, methylome and proteome, studying “omics” outcomes may lead to an improved understanding of disease processes. In particular, joint analysis of multi-omics data may enhance our knowledge of how genetic effects on these outcomes are coordinated in a multi-level molecular system to contribute to disease susceptibility. Since genetic effects on molecular phenotypes may further depend on tissue, cell type, or other conditions, the scientific community would benefit from continued development of methods to integrate multi- omics data across conditions or contexts. However, the large scale of the data coupled with unknown correlation structures across features or conditions makes such analyses challenging. In this project, we propose efficient methods to integrate summary statistics from multiple studies of genetic effects on complex and omics phenotypes. To improve upon existing multi-omics integrative approaches that take summary statistics as input, we expand joint analyses to more than three data types or conditions, and allow the sets of statistics to come from overlapping samples. Preliminary results presented in the application demonstrate that the proposed methods are computationally feasible and produce results that are consistent with current biological knowledge. Proposed applications of the methods have the potential to identify novel associations or provide new evidence for known associations between omics features and cancer risk. The success of this work will provide flexible methods and computational tools that can be applied to other diseases and settings.