基因-基因(环境)交互作用分析方法的比较  

Analytical Methods Comparison for Gene-gene (Environment) Interactions

作　　者：袁芳[1] 刘盼盼[1] 徐进[1] 费丽娟[1] 郝玲妹 邱旭君 张莉娜[1] 

机构地区：[1]宁波大学医学院,浙江宁波315211 [2]宁波市第七医院,浙江宁波315200

出　　处：《宁波大学学报（理工版）》2012年第4期115-119,共5页Journal of Ningbo University：Natural Science and Engineering Edition

基　　金：浙江省教育厅重点科研项目(Z201017918);宁波市自然科学基金(2011A610037)

摘　　要：对复杂疾病病因研究中基因-基因(环境)交互作用的3种分析方法进行了比较,剖析了它们的适用条件和优缺点.结果表明:叉生分析简单易行,但只适用于分析单个遗传和单个环境因素的交互作用;Logistic回归易解释交互作用的流行病学意义且能很好地分析主效应,但在分析高阶交互作用时存在局限性;多因子降维法无需指定特定的遗传模式且对高维数据敏感,但无法估计主效应.鉴于这3种方法在分析交互作用时各有其优点,三者联合应用于交互作用分析效果更佳.Three methods of analyzing the gene-gene （environment） interactions are compared in the etiology research of complex diseases to analyze their applicability conditions and the advantages and disadvantages. It shows that crossover analysis is simple and easy to apply, but only applicable for analyzing interactions of single genetic factor and single environment factor. Logistic regression is straight-forward in explaining the epidemiological significance of interaction and performs well in analyzing the main effects, but still has limitations in analyzing higher order interactions. Multifactor dimensionality reduction renders a model-free method and is sensitive to high dimensional data, but it is short of accuracy in estimating the main effects. Considering the advantages identified with each of these three methods in analyzing interaction, the author makes some efforts in this paper to integrate these methods aiming at improving effectiveness interaction analysis.

关 键 词：叉生分析 LOGISTIC回归 多因子降维法 交互作用 

分 类 号：R181[医药卫生—流行病学] R195.1[医药卫生—公共卫生与预防医学]