Mendelian randomization (MR) approach follows the Mendel's law of inheritance, which is called "Parental alleles randomly assigned to the offspring", and refers to use genetic variants as an instrumental variable to develop causal inference between the exposure factor and the outcome from observational study. In recent years, with the rapid development of genome-wide association study (GWAS) and various omics data,the disclosure of a large number of aggregated data provides an opportunity for the wide application of MR approach in causal inference. We introduce three methods widely used in MR and then apply them to explore causal relationship between blood metabolites and depressive. The advantages and disadvantages of three methods in causal inference are compared in order to provide reference for the application of MR in observational studies.
孟德尔随机化方法，遵循孟德尔遗传定律，"亲代等位基因随机分配给子代"，以遗传变异作为工具变量，以此推断观察性研究中暴露因素与研究结局的因果关联。近年来，随着全基因组关联研究（GWAS）及各种组学数据的飞速发展，大量汇总数据的公开，为孟德尔随机化方法在因果推断中的广泛应用提供契机。本文针对目前常用的三种孟德尔随机化方法进行阐述，并将其应用于探索血液代谢物与抑郁症的因果关联，比较三种方法在因果推断的优缺点，以期为孟德尔随机化在观察性研究中的应用提供参考。.
Keywords: Causal inference; Genetic variation; Mendelian randomization analysis; Methods.