Mendelian randomization is an instrumental variable technique to estimate the causal effects of risk factors on an outcome. Each genetic variant fulfilling certain conditions can be used as an instrument and leveraged to provide such an estimate. In cases where distinct processes underlie a causal relationship, the resulting estimates may differ substantially, and hence such discrepancies could be detected.
An article in this issue, by Verkouter et al., proposes an approach to define groups of instruments a priori—based on their potential tissue-specific mechanism—and examine whether tissue-grouped causal effects show evidence of divergent mechanisms