基于Adaptive Lasso的两阶段全基因组关联分析方法  

ALGWAS:two-stage Adaptive Lasso-based genome-wide association study

作　　者：杨文宇 吴成秀 肖英杰 严建兵 YANG Wen-Yu;WU Cheng-Xiu;XIAO Ying-Jie;YAN Jian-Bing(National Key Laboratory of Crop Genetic Improvement,Huazhong Agricultural University,Wuhan 430070,Hubei,China;College of Science,Huazhong Agricultural University,Wuhan 430070,Hubei,China;Hubei Hongshan Laboratory,Wuhan 430070,Hubei,China)

机构地区：[1]作物遗传改良全国重点实验室,湖北武汉430070 [2]华中农业大学理学院,湖北武汉430070 [3]湖北洪山实验室,湖北武汉430070

出　　处：《作物学报》2023年第9期2321-2330,共10页Acta Agronomica Sinica

基　　金：国家自然科学基金项目(32201855,32122066)资助。

摘　　要：作为进行全基因组关联分析的主流方法,混合线性模型类方法得到了广泛的应用。但是,现有方法仍存在检测功效不高的问题。本文提出一种基于AdaptiveLasso的2阶段全基因组关联分析方法(two-stage Adaptive Lasso-based genome-wide association analysis, ALGWAS),该方法在第1阶段通过变量选择方法 Adaptive Lasso筛选出与目标性状相关联的单核苷酸多态性位点(single nucleotide polymorphism, SNP),第2阶段将第1阶段筛选出的SNP作为协变量放入线性模型中进行全基因组扫描。在模拟实验中,ALGWAS方法与3种常用的全基因组关联分析方法fastGWA、GEMMA和EMMAX相比具有最高的检测功效,同时具有较低的错误发现率(falsediscoveryrate,FDR)。将以上4种方法应用到包含1341份材料的玉米CUBIC (Complete-diallel plus Unbalanced Breeding-like Inter-Cross)群体的全基因组关联分析中,ALGWAS方法可检测到与开花期相关基因ZmMADS69、ZmMADS15/31、ZmZCN8和ZmRAP2.7,与株高相关基因ZmBRD1和ZmBR2,与产量相关基因ZmUB2、ZmKRN2和ZmCLE7等,而其他3种常用的全基因组关联分析方法检测功效较低。本研究提出了一种非混合线性模型类的全基因组关联分析方法,对解析微效多基因决定的复杂遗传性状具有更高的检测效率,为基因挖掘提供了新的途径。As mainstream methods for genome-wide association analysis,mixed linear model methods have been widely used.However,the existing methods still have the problem of low detection power.In this study,a two-stage Adaptive Lasso-based genome-wide association analysis(ALGWAS)method was proposed.In the first stage,single nucleotide polymorphism(SNP)associated with target traits were screened by Adaptive Lasso,a variable selection method.In the second stage,SNPs selected from the first stage were put into the linear model as the covariates for genome-wide scanning.Compared with fastGWA,GEMMA and EMMAX,the ALGWAS method had the highest detection power and lower false discovery rate(FDR)in the simulation experiments.The above four methods were applied to genome-wide association analysis of Complete-diallel plus Unbalanced Breeding-like Inter-Cross(CUBIC)population of 1341 individuals in maize.ALGWAS method can detect the genes(ZmMADS69,ZmMADS15/31,ZmZCN8,and ZmRAP2.7)related to days to tasseling,the genes(ZmBRD1 and ZmBR2)related to plant height,and the genes(ZmUB2,ZmKRN2,and ZmCLE7)related to yield,while the other three commonly used genome-wide association analysis methods had low detection efficiency.In this study,a non-mixed linear model class of genome-wide association analysis method was proposed,which had higher detection advantage for microeffect polygenes and provided a new way for genetic analysis of complex traits.

关 键 词：玉米 全基因组关联分析 变量选择 Adaptive Lasso 

分 类 号：S513[农业科学—作物学]