Predicting Phenotype by Using Transcriptomic Alteration as Endophenotype

项目摘要 对人类复杂特征或疾病的遗传结构的现代研究通常分为三种 关联关系的设计：遗传变异与疾病之间的关联， 遗传变异和表达之间的关系（例如表达数量性状基因座，eQTL），以及 基因表达和疾病之间的联系许多有希望的发现被发现，包括数千个单一的 核苷酸多态性与常见疾病有关。虽然这些发现为我们提供了 对常见疾病的遗传结构和疾病之间的共同遗传性的宝贵见解， 缺少的是机制，包括确切的因果变量，其影响的方向，以及 事件的顺序，这构成了我们希望通过本研究解决的基本假设。 提议在许多最近发现的启发下， 遗传变异位于调控区，在这个建议中，我们联合收割机结合生物信息学，统计 遗传学、精准医学、表型和电子病历（EMR）数据挖掘， 最大限度地利用来自基因型和表达的调控信息的新型分析策略， 目的是利用转录组学改变和DNA变异来预测表型。我们提出以下三点 主要目标。(1)将遗传学和遗传学相结合，建立一个统一的遗传模型， 转录组学关联从ENCODE生成的功能和监管注释数据， FANTOM 5、GENCODE、表观基因组路线图和GTEx将被有效地结合起来，以推断出 重要的内表型，基因决定的表达成分，用于更好地预测 表型或疾病结果。(2)开发基于最大似然的链接检验和表型特异性 调控网络方法解析基因介导的基因型-表型因果关系 表情(3)广泛评估精神分裂症的方法并将其应用于广泛的表型 使用范德比尔特生物库（BioVU）基因型和链接的电子医疗数据。基于我们之前的 研究和强大的初步数据，这一建议是及时研究人类遗传结构 复杂的疾病和性状通过解剖遗传成分贡献的调节作用的变异 在基因表达水平上。这是非常重要的，因为它解决了强大的限制，在许多 全基因组关联研究（GWAS）和下一代测序（NGS）用于推断因果关系， 在精准医疗的新兴领域的转化潜力。这个项目的顺利完成 不仅将促进我们对精神分裂症遗传成分的理解， 表型或临床结果，而且还为公众提供了有用的方法和工具， 通过基因组和医学信息的联系研究表型的遗传结构。

Project Summary Modern studies of the genetic architecture underlying human complex traits or diseases generally fall into three designs of association relationship: the association between genetic variants and disease, the association between genetic variants and expression (e.g. expression quantitative trait loci, eQTL), and the association between gene expression and disease. Many promising findings are discovered, including thousands of single nucleotide polymorphisms found to be associated with common diseases. While these findings provide us with valuable insights into the genetic architecture of common diseases and the shared heritability among diseases, what missing are the mechanisms, including the exact causal variants, the direction of their effects, and the orders of events, which forms the foundational hypothesis that we would like to solve through the studies in this proposal. With the inspiration of many recent discoveries that a substantial fraction of the disease-associated genetic variants is located in regulatory regions, in this proposal, we combine bioinformatics, statistical genetics, precision medicine, and phenotype and electronic medical record (EMR) data mining to develop novel analytical strategies that maximally leverage regulatory information from both genotype and expression, aiming to predict phenotype using transcriptomic alteration with DNA variation. We propose the following three major aims. (1) To build a unified genetic model for the prediction of phenotype by combining genetic and transcriptomic associations. Functional and regulatory annotation data generated from the ENCODE, FANTOM5, GENCODE, the Epigenomic Roadmap, and GTEx will be effectively incorporated to infer an important endophenotype, the genetically determined expression component, for better prediction of phenotype or disease outcome. (2) To develop a maximum likelihood based link test and a phenotype-specific regulatory network approach to resolve genotype-phenotype causality relationships mediated by gene expression. (3) To extensively evaluate the approaches in schizophrenia and apply them to broad phenotypes using the Vanderbilt biobank (BioVU) genotype and linked electronic medical data. Building on our previous studies and strong preliminary data, this proposal is timely for studying the genetic architecture in human complex diseases and traits by dissecting the genetic components contributed from regulatory roles of variants at the gene expression level. It is highly significant because it tackles the strong limitations in numerous genome-wide association studies (GWAS) and next-generation sequencing (NGS) for inferring causality and translational potentials in the emerging fields of precision medicine. The successful completion of this project will not only advance our understanding of genetic components in schizophrenia and a broad spectrum of phenotypes or clinical outcomes, but also provide useful methods and tools to the public community for studying genetic architecture of phenotype via the linkage of genomic and medical information.