Epistatic and Cross Tissue Analysis for Human Gene Expression Traits

描述(由申请人提供)： 人类基因表达特征的上位性和跨组织分析基因组广泛关联研究(GWAS)提供了前所未有的发现速度，将DNA变异与常见的人类疾病联系起来。然而，在大多数情况下，这些SNP如何影响人类疾病尚不清楚。基因表达是SNPs与疾病表型之间的中介。最大限度地利用在多个组织上收集的人类队列中的基因表达和遗传变异信息的方法显示出巨大的希望，不仅可以表征疾病的遗传结构，而且还可以表征定义疾病的分子网络。这项应用的长期目标是开发和实施新的统计方法，以确定影响个人对复杂表型(如疾病)的易感性的基因网络。在这里，我们提出了几个模型开发，它们不仅增强了我们在单个或多个组织中检测eQTL的能力，而且在定义与疾病相关的生物学过程的分子网络环境中识别eQTL以及eQTL之间的相互作用：(1)将开发一种同时对所有基因和所有标记的总分布进行建模的贝叶斯建模方法。我们方法的优势将是它能够在边际效应较弱的情况下以高功率检测上位性，解决了所有其他eQTL作图方法的一个关键弱点。(2)将开发一种基于可能性的方法来推断因果关系，该方法还包括转录因子结合位点信息。(3)将开发一种将不同组织中的子网络与疾病联系起来的方法。此外，还将开发一种分析组织之间因果关系/反应关系的方法。(4)所提出的方法将通过模拟，更重要的是，在我们生成的多组织鼠和人类队列数据上得到广泛的验证。所有方法都将在用户友好的软件中实施，并向科学界提供。 公共卫生相关性： 人类基因表达性状的上位性和跨组织分析基因表达是SNPs和疾病表型之间的中介。最大限度地利用在多个组织上收集的人类队列中的基因表达和遗传变异信息的方法显示出巨大的希望，不仅可以表征疾病的遗传结构，而且还可以表征定义疾病的分子网络。我们提出了几个模型开发，不仅增强了我们在单个或多个组织中检测eQTL的能力，而且在定义与疾病相关的生物过程的分子网络环境中识别eQTL以及eQTL之间的相互作用。

DESCRIPTION (provided by applicant): Epistatic and cross-tissue analysis for human gene expression traits Genome wide association studies (GWAS) have delivered unprecedented rates of discovery associating variations in DNA with common human diseases. However, how these SNPs affect human diseases are not clear in most cases. Gene expression is the intermediate between SNPs and disease phenotypes. Methods to maximally leverage gene expression and genetic variation information collected in human cohorts over multiple tissues show great promise for characterizing not only the genetic architecture of disease but the molecule networks that define disease. The long-term goal of this application is to develop and implement novel statistical methods to identify networks of genes affecting an individual's susceptibility to complex phenotypes like disease. Here, we propose several model developments that not only enhance our power to detect eQTL in single or multiple tissues, but that identify eQTL and interactions among eQTL in the molecular network contexts that define biological processes associated with disease: (1) A Bayesian modeling approach that simultaneously models the total distribution of all genes and all markers will be developed. The strength of our approach will be its ability to detect epistasis with high power when the marginal effects are weak, addressing a key weakness of all other eQTL mapping methods. (2) A likelihood based approach for inferring causal relationships that also incorporates transcription factor binding site information will be developed. (3) An approach for linking subnetworks in different tissues to diseases will be developed. Also a method to dissect causal/ reactive relationships between tissues will be developed. (4) The proposed methods will be extensively validated via simulations and, more importantly, on multi-tissue mouse and human cohort data we have generated. All methods will be implemented in user-friendly software and made available to the scientific community. PUBLIC HEALTH RELEVANCE: Epistatic and cross-tissue analysis for human gene expression traits Gene expression is the intermediate between SNPs and disease phenotypes. Methods to maximimally leverage gene expression and genetic variation information collected in human cohorts over multiple tissues show great promise for characterizing not only the genetic architecture of disease but the molecule networks that define disease. We propose several model developments that not only enhance our power to detect eQTL in single or multiple tissues, but that identify eQTL and interactions among eQTL in the molecular network contexts that define biological processes associated with disease.