Integrative modeling of gene expression into GWAS of glioma

描述（申请人提供）：胶质瘤占所有恶性脑肿瘤的80%，由于预后不良，给临床和公共卫生造成巨大负担。鉴于神经胶质瘤的病因尚不清楚，了解遗传易感性可能会为揭示这种致命癌症背后的生物学机制提供新的见解和机会。全基因组关联研究（GWAS）是研究单核苷酸多态性（snp）与疾病之间关系的一种流行方法。另一方面，研究遗传变异如何改变组织中基因表达的研究（即数量性状位点（quantitative trait loci, eQTL）研究）侧重于分子数量性状。由于传统的GWAS分析由于其不可知的方法而受到功率损失，因此需要新的策略来识别额外的和科学上有意义的神经胶质瘤风险易感位点。因此，我们建议整合eQTL研究，以便在不同受试者之间进行eQTL研究和GWAS研究时，更有力地测试SNP对GWAS疾病的影响。通过SNPs和基因表达对疾病风险的联合影响的回归模型，我们开发了一种有效的测试程序，用于测试基因或途径中设置的eQTL SNP的总体影响，并说明了其在数值模拟研究和哮喘研究中的实用性。我们将以三个具体的研究目标进行综合分析。在Aim 1中，我们将使用从神经正常受试者和胶质瘤患者的脑组织中收集的全基因组SNP和表达数据进行eQTL分析。在Aim 2中，我们将首先在Aim 1中鉴定的基因或免疫调节途径中形成eQTL SNP集，然后使用公开可用的胶质瘤风险GWAS数据进行eQTL SNP集分析，以研究eQTL SNP与胶质瘤风险的关联。在Aim 3中，我们将对Aim 2中鉴定的一个基因的snp集及其表达值进行综合分析，以评估基因对胶质瘤风险的影响，无论是通过eQTL snp的直接影响还是通过基因表达介导的间接影响。本项目的目标是建立一个进行和分析GWAS的新框架，并确定胶质瘤的新易感位点。由于不同的基因组数据（即snp和基因表达）被整合到分析中，我们期望在检测疾病驱动易感性位点方面比单核苷酸多态性分析具有更强的统计能力。此外，我们的eQTL整合方法的结果也将比传统的不可知GWAS更具生物学意义和可解释性，因为eQTL snp更有可能是功能性的，而且eQTL对疾病风险的影响是明确建模的。

DESCRIPTION (provided by applicant): Gliomas account for 80% of all malignant brain tumors and produce enormous clinical and public health burden due to their poor prognosis. Given the lack of known causes for glioma, understanding genetic susceptibility might provide new insights and opportunities for progress in unraveling the biological mechanisms behind this fatal cancer. Genome-wide association studies (GWAS) constitute a popular approach for investigating the association of single nucleotide polymorphisms (SNPs) with disease. On the other hand, studies examining how genetic variants modify gene expression in tissue (i.e., quantitative trait loci (eQTL) studies) focus on the molecular quantitative trait. As the traditionl GWAS analysis is subject to power loss due to its agnostic approach, new strategies are required to identify additional and scientifically meaningful susceptibility loci of glioma risk. Hre we propose to integrate eQTL studies to more powerfully test the SNP effect on disease in GWAS when eQTL studies and GWAS are conducted among different subjects. With a regression model for the joint effect of SNPs and gene expression on disease risk, we have developed an efficient testing procedure for the overall effect of an eQTL SNP set in a gene or a pathway and illustrated its utility in numerical simulation studies and an asthma study. We will pursue the integrated analysis with three specific study aims. In Aim 1, we will conduct eQTL analyses using genome-wide SNP and expression data collected from brain tissue obtained from neurologically normal subjects and glioma patients. In Aim 2, we will first form eQTL SNP sets in a gene or an immunomodulating pathway identified from Aim 1 and then conduct eQTL SNP-set analyses to investigate the association of eQTL SNPs with the risk of glioma using the publicly available GWAS data of the glioma risk. In Aim 3, we will perform integrated analyses of the SNP-set of a gene identified in Aim 2 and its expression value to assess the gene-based effect on the risk of glioma, either through a direct effect of eQTL SNPs or an indirect effect mediated by gene expression. The goal of this project is to build a new framework of conducting and analyzing a GWAS and identify new susceptibility loci for glioma. As different genomic data (i.e., SNPs and gene expression) are integrated in the analysis, we would expect a more statistical power to detect the disease-driving susceptibility loci than the SNP-only analysis. Furthermore, the results from our eQTL-integrated approach will also be more biologically meaningful and interpretable than the conventional agnostic GWAS because eQTL SNPs are more likely to be functional and the eQTL effect on the disease risk is explicitly modeled.