Mendelian randomization for causal inference accounting for pleiotropy and sample structure using genome-wide summary...

Area of Science:

• Genetics
• Statistical Genetics
• Epidemiology

Background:

• Mendelian randomization (MR) infers causality using genome-wide association study (GWAS) summary statistics.
• Existing MR methods often make strong assumptions, leading to false positives, primarily by not fully accounting for pleiotropy.
• Sample structure, including population stratification, cryptic relatedness, and sample overlap, is identified as a significant, underappreciated confounding factor in MR.

Purpose of the Study:

• To develop a unified MR approach that simultaneously accounts for both pleiotropy and sample structure.
• To enhance statistical power in MR analyses by enabling the inclusion of more genetic variants with moderate effects as instrument variables (IVs).
• To improve the reliability of causal inference, particularly for complex and highly polygenic traits.

Main Methods:

• Proposed MR-APSS, a novel MR method leveraging genome-wide information.
• MR-APSS simultaneously adjusts for confounding due to pleiotropy and sample structure.
• Evaluated MR-APSS through comprehensive simulations, negative controls, and application to diverse complex traits.

Main Results:

• MR-APSS demonstrated improved reliability in identifying plausible causal relationships compared to existing methods.
• The method effectively mitigates confounding from both pleiotropy and sample structure.
• MR-APSS showed robust performance for highly polygenic traits with weak instrument variables, where other methods are vulnerable.

Conclusions:

• MR-APSS offers a more robust framework for causal inference in genetic epidemiology.
• Accounting for sample structure alongside pleiotropy is crucial for accurate MR findings.
• The developed method enhances the utility of MR for studying complex traits.