GENIUS-MAWII: for robust Mendelian randomization with many weak invalid instruments

Area of Science:

• Epidemiology
• Statistical Genetics
• Bioinformatics

Background:

• Mendelian randomization (MR) uses genetic variants as instrumental variables to infer causal relationships.
• Key challenges in MR include weak instruments and horizontal pleiotropy, which can bias results.
• Existing methods struggle to address both issues simultaneously.

Purpose of the Study:

• To propose a novel Mendelian randomization method, GENIUS-MAny Weak Invalid IV, that addresses multiple weak and invalid instruments and widespread horizontal pleiotropy.
• To develop a robust statistical framework for causal inference in the presence of these common MR challenges.
• To provide practical tools for assessing the validity and reliability of MR analyses.

Main Methods:

• The proposed method, GENIUS-MAny Weak Invalid IV, utilizes the heteroscedasticity of the exposure to identify the treatment effect.
• It involves deriving influence functions for the treatment effect and constructing a continuous updating estimator.
• Novel semiparametric theory is developed to establish asymptotic properties under a many weak invalid instruments regime.

Main Results:

• The study establishes the asymptotic properties of the proposed continuous updating estimator under challenging MR conditions.
• A novel measure for weak identification is introduced.
• An overidentification test and a graphical diagnostic tool are provided to aid MR analysis.

Conclusions:

• GENIUS-MAny Weak Invalid IV offers a statistically rigorous approach to Mendelian randomization when faced with multiple weak and invalid instruments and horizontal pleiotropy.
• The developed methods and tools enhance the reliability of causal inference from genetic association studies.
• This work contributes to advancing robust causal inference methodologies in genetic epidemiology.