MRCIP: a robust Mendelian randomization method accounting for correlated and idiosyncratic pleiotropy

Area of Science:

• Genetics
• Statistical Genetics
• Epidemiology

Background:

• Mendelian randomization (MR) uses genetic variants as instrumental variables (IVs) to estimate causal effects.
• Unaccounted pleiotropy, including correlated and idiosyncratic forms, can bias MR estimates.
• Existing methods struggle to simultaneously address both types of pleiotropy.

Purpose of the Study:

• To develop a novel MR approach, MRCIP, for simultaneously accounting for correlated and idiosyncratic pleiotropy.
• To improve the accuracy of causal effect estimation in the presence of complex pleiotropic effects.
• To provide a statistical framework for quantifying and testing correlated pleiotropy.

Main Methods:

• Development of a random-effect model to explicitly capture correlated pleiotropy.
• Introduction of a novel weighting scheme to manage idiosyncratic pleiotropy.
• Parameter estimation via maximizing a weighted likelihood function using the PRW-EM algorithm.

Main Results:

• MRCIP demonstrates improved performance compared to existing methods in extensive simulation studies.
• The method successfully estimates the degree of correlated pleiotropy and allows for its statistical testing.
• Application to two real datasets highlights the practical utility of MRCIP.

Conclusions:

• MRCIP offers a robust solution for Mendelian randomization analyses affected by both correlated and idiosyncratic pleiotropy.
• The developed method enhances the reliability of causal inference from genetic association studies.
• The MRCIP R package is publicly available for broader research application.