MR-GRAPPLE is a method that uses instrumental variables (IVs) that are strongly and weakly associated with the exposure to identify possible pleiotropic pathways and obtain an unbiased causal effect estimate.
MR-GRAPPLE uses IVs that are associated with the exposure at varying p-value thresholds to detect possible pleiotropic pathways that may bias the causal effect estimate by assessing the stability of the causal effect over various thresholds. The method also determines the causal direction and can be extended to perform multivariable MR to adjust for potential confounding factors. The method increases statistical power in situations where there may be one or few IVs strongly associated with the exposure or where polygenic traits have many IVs with a variety of strengths and can provide valid inference that avoids weak instrument bias as well as bias from horizontal pleiotropy.