调整协变量的多位点关联分析方法及其在酗酒数据中的应用  

A Multi-Locus Association Study Method Adjusting for Covariates and Its Application to COGA Dataset

作　　者：邓舒方 王昱泉 胡跃清 DENG Shufang;WANG Yuquan;HU Yueqing(Department of Biostatistics and Computational Biology, School of Life Sciences, Fudan University, Shanghai 200433, China)

机构地区：[1]复旦大学生命科学学院生物统计学与计算生物学系,上海200433

出　　处：《复旦学报（自然科学版）》2021年第6期768-778,788,共12页Journal of Fudan University：Natural Science

基　　金：国家自然科学基金(11971117)。

摘　　要：复杂疾病的发生与个体的遗传变异和环境协变量密切相关.在进行遗传位点与性状的关联分析时,环境协变量起着重要作用.如果处理它们的方法不当,可能会带来分析结果的偏差或是检验功效的降低.针对多位点遗传变异信息,我们在整合连锁不平衡信息的检验方法SLIDE的基础上,提出了对基因型用倾向得分进行逆概率加权的调整协变量关联分析方法SLIDE_(a).随机模拟结果表明,当协变量既与基因型相关又与性状相关时,SLIDE_(a)可以有效控制住第一类错误率.同时,当致病位点包含常见变异,多位点对性状效应方向不一致或位点与协变量存在交互作用时,SLIDE_(a)具有优越性.随后我们将SLIDE_(a)应用于COGA酗酒相关数据分析中,找到了OPA3,OXTR等数个与酒精成瘾性相关的基因,这展示了SLIDE_(a)的有效性并为后续的生物学研究提供理论指引.The human complex diseases are closely related to both genetic variation of a subject and environmental covariates.In association study of genotypes and traits,covariates play an essential role,and inappropriate utilization of these covariates may lead to bias or reduced power in statistical analysis.On the basis of the test statistic incorporating linkage disequilibrium(SLIDE),we propose SLIDE_(a),using inverse probability weighted propensity score on multi-locus genotypic data.Simulation results demonstrate that SLIDE_(a) can control typeⅠerror rates well when there is correlation between covariates,genotype and trait.In the meantime,if the causal variants are common,and their effects are different in direction,or the interaction of variants and covariates has influence on the trait,SLIDE_(a) exhibits superiority over the traditional methods.At last,we apply SLIDE_(a) to COGA dataset and find several genes related to alcohol addiction,including OPA3 and OXTR.This further illustrates the validity of SLIDE_(a) and provides theoretical guidelines to future biological analysis.

关 键 词：多位点 关联分析 调整协变量 倾向得分 

分 类 号：O212.2[理学—概率论与数理统计]