Two-sample bi-directional causality between two traits with some invalid IVs in both directions using GWAS summary...

Area of Science:

• Genetic Epidemiology
• Statistical Genetics
• Causal Inference

Background:

• Mendelian randomization (MR) assesses causality using genetic variants as instrumental variables (IVs).
• Traditional MR assumes uni-directional causality, while bi-directional MR seeks true causal direction.
• Pleiotropy-induced invalid IVs challenge both uni-directional and bi-directional MR analyses.

Purpose of the Study:

• To introduce a new bi-directional MR method, Bidir-SW, addressing challenges of invalid IVs.
• To account for potential bi-directional causality even with common invalid IVs.
• To provide a robust framework for genetic epidemiology studies.

Main Methods:

• Developed Bidir-SW, a bi-directional MR method with stepwise selection.
• Utilized public genome-wide association study (GWAS) datasets.
• Employed model selection criteria for iterative identification of invalid IVs.

Main Results:

• Bidir-SW demonstrated superior performance over traditional MR methods (MR-Egger, IVW) in simulations with uncorrelated SNPs.
• Simulations confirmed Bidir-SW's effectiveness compared to transcriptome-wide association study (TWAS).
• Applied Bidir-SW to CRP levels and BMI, revealing bi-directional relationships and identifying causal protein biomarkers.

Conclusions:

• Bidir-SW is a powerful tool for bi-directional MR and TWAS.
• The method offers a valuable framework for future genetic epidemiology research.
• Bidir-SW effectively handles invalid genetic instruments in causal inference.