Towards more plausible point-identifying assumptions in two-sample Mendelian randomization
两样本孟德尔随机化中更具可信度的点识别假设
Two-sample Mendelian randomization (MR) is a widely applied methodology in epidemiology. In two-sample MR, summary data (typically, regression coefficients and standard errors) quantifying the association between multiple genetic variants and the exposure and the outcome are used in an instrumental variable framework aimed at estimating the causal effect of the exposure on the outcome. Most two-sample MR methods were developed under data-generating models where the…