Multi-omics Gene Network Identification (Project 4)

项目摘要/摘要 项目4的目标是发现阿片类药物背后的神经生物学可解释的基因网络 成瘾(OA)，但在其他方面被传统统计方法遗漏。为了实现这一目标，我们将 应用我们的多组学、多方法框架--基因网络识别和整合(GNetII)-- 通过利用人类和啮齿动物模型中的P50数据，识别与OA相关的基因网络。GNetII 包括全基因组的上位性，可解释的人工智能，基因网络的构建，以及 证据(LOE)方法。这些基石方法实现了多个级别的个人级别的集成 数据，我们将专门整合220,722中的大规模全基因组关联研究(GWAS)数据 来自项目1和协同核心的活的受试者，基因调控数据(RNA测序，DNA 甲基化、染色质免疫沉淀测序和变异基因类型)来自多种成瘾- 相关脑组织(包括前额叶皮质、伏隔核和杏仁核) 对照项目2中的死者(死亡个体)，实验小鼠和大鼠模型结果 项目3，以及其他公共组学数据。 在美国，骨关节炎是可预防的发病率和死亡率的主要原因，困扰着 空前数量的美国成年人和年轻人。骨性关节炎具有很高的遗传性(~54%)。然而，很少有遗传基因座 被最终确定为骨性关节炎和相关结果，以及常见的基于遗传变异的遗传性 仅解释了办公自动化中17%的差异。我们假设上位性(即变异或基因的相互作用) 造成了缺失的遗传性。将大数据科学方法应用于大规模GWAs、GENE 脑组织中的调节数据和跨物种数据将揭示以前未被发现的关系，并增加 了解成瘾背后的神经生物学知识。我们提出了以下具体目标： 目的1：通过全基因组上位性效应构建骨性关节炎相关基因网络。 目的2：利用死后人脑数据建立多组学网络。 目的3：整合跨物种的网络以发现具有多个LOE的OA相关基因网络。 将对橡树岭强大的高性能计算体系结构进行分析 这将极大地提高有神经生物学意义的发现的可能性。我们的 这项研究将捕捉整个基因组的复杂网络，以发现以前未知的基因，并将有所帮助 解释从P50和P50出现的遗传位点的神经生物学基础 更广阔的领域。

PROJECT SUMMARY/ABSTRACT The goal of Project 4 is to discover neurobiologically interpretable gene networks that underlie opioid addiction (OA) but are otherwise missed by traditional statistical approaches. To achieve this goal, we will apply our multi-omics, multi-method framework—Gene Network Identification and Integration (GNetII)—to identify OA-associated gene networks, by capitalizing on P50 data across human and rodent models. GNetII includes genome-wide epistasis, explainable artificial intelligence, gene network construction, and lines-of- evidence (LOE) methods. These cornerstone methods enable integration of multiple levels of individual-level data, and we will specifically integrate large-scale genome-wide association study (GWAS) data in 220,722 living subjects from Project 1 and the Synergy Core, gene regulation data (RNA-sequencing, DNA methylation, chromatin immunoprecipitation sequencing, and variant genotypes) from multiple addiction- relevant brain tissues (including prefrontal cortex, nucleus accumbens, and amygdala) from 641 OA case and control decedents (deceased individuals) from Project 2, experimental mouse and rat model results from Project 3, and additional public omics data. OA is a leading cause of preventable morbidity and mortality in the United States, afflicting an unprecedented number of U.S. adults and youth. OA is highly heritable (~54%). Yet, few genetic loci have been conclusively identified for OA and related outcomes, and common genetic variant-based heritability explains only 17% of the variance in OA. We hypothesize that epistasis (i.e., interaction of variants or genes) contributes to the missing heritability. Applying big data science methods to large-scale GWAS, gene regulation data in brain tissue, and cross-species data will reveal previously undetected relationships and add knowledge of the neurobiology underlying addiction. We propose the following specific aims: Aim 1: Build OA-associated gene networks via genome-wide epistasis. Aim 2: Build multi-omics networks using postmortem human brain data. Aim 3: Integrate networks across species to find OA-associated gene networks with multiple LOE. Analyses will be performed on the powerful high performance computing architectures at the Oak Ridge National Laboratory which will greatly improve the likelihood of neurobiologically meaningful discoveries. Our study will capture complex networks across the genome to find previously unknown genes and will help explain the neurobiological underpinnings for the genetic loci that emerge from across the P50 and the broader field.