[Mendelian randomization approach, used for causal inferences].

Mendelian randomization (MR) approach is based on the Mendelian genetic law, which is called "Parental alleles that randomly assigned to the offspring" . MR refers to the use of genetic variants to develop causal inferences from observational data, if the variant genotype is associated with the phenotype and the variant genotype associated with the risk of disease of interest through the phenotype. Hence, the genotype can be used as Instrumental Variable (IV) to infer the causal relation between the phenotype and the risk of diseases. In recent years, MR approach is widely used in causal inference between the exposure factors and the risks of disease, along with the rapid development of statistical methods, big datasets of GWAS, epigenetics and the various "omics" techniques. This paper provides an overview of the MR strategies and addresses the related assumptions and implications, with reliability and limitations included.