MendelianRandomization v0.9.0: updates to an R package for performing Mendelian randomization analyses using...

Area of Science:

• Biostatistics
• Genetic Epidemiology
• Statistical Genetics

Background:

• Mendelian randomization (MR) is a powerful method for causal inference using genetic variants as instrumental variables.
• Summarized data analysis in MR is efficient but requires robust methods to handle biases.
• Previous versions of the MendelianRandomization package provided foundational tools for MR.

Purpose of the Study:

• To describe enhancements and new functionalities in the MendelianRandomization R package since version 0.5.0.
• To introduce robust methods for addressing bias in Mendelian randomization analyses.
• To provide tools for handling correlated genetic variants and assessing instrument strength.

Main Methods:

• Implementation of new robust Mendelian randomization methods to mitigate bias from weak instruments, winner's curse, and pleiotropic variants.
• Development of dimension reduction techniques for Mendelian randomization with large numbers of correlated genetic variants.
• Inclusion of functions for calculating first-stage F statistics (instrument strength) in univariable and multivariable MR with both correlated and uncorrelated variants.

Main Results:

• The updated package offers improved robustness against common sources of bias in Mendelian randomization.
• New methods facilitate the application of Mendelian randomization in the presence of correlated genetic variants.
• Enhanced functions provide comprehensive assessment of instrument strength in various MR settings.

Conclusions:

• The enhanced MendelianRandomization package provides advanced, robust tools for causal inference using summarized genetic data.
• The new functionalities improve the reliability and applicability of Mendelian randomization studies, particularly with complex genetic data.
• The package supports researchers in conducting rigorous causal inference by addressing key methodological challenges.