Mendelian randomization accounting for correlated and uncorrelated pleiotropic effects using genome-wide summary...

Area of Science:

• Genetics
• Statistical Genetics
• Epidemiology

Background:

• Mendelian randomization (MR) utilizes genetic variants to infer causal relationships between traits.
• The increasing availability of genome-wide association studies (GWAS) expands opportunities for MR applications.
• Existing MR methods are susceptible to false positives, particularly from horizontal pleiotropy, where genetic variants influence multiple traits through shared pathways.

Purpose of the Study:

• To introduce a novel MR method, Causal Analysis Using Summary Effect estimates (CAUSE), designed to account for both correlated and uncorrelated horizontal pleiotropy.
• To evaluate the performance of CAUSE in mitigating false positives caused by correlated horizontal pleiotropy compared to existing methods.
• To apply CAUSE to real-world GWAS data to identify robust causal relationships.

Main Methods:

• Development of the CAUSE method for Mendelian randomization analysis.
• Simulation studies to assess the accuracy of CAUSE in the presence of correlated horizontal pleiotropy.
• Application of CAUSE to summary statistics from recent genome-wide association studies.

Main Results:

• Simulations demonstrate that CAUSE significantly reduces false positives arising from correlated horizontal pleiotropy compared to other MR approaches.
• Application of CAUSE to GWAS data identified several causally supported relationships with strong existing literature evidence.
• The method successfully avoided identifying numerous spurious relationships that might be suggested by alternative methods.

Conclusions:

• Correlated horizontal pleiotropy is a common issue in genetic association studies, potentially leading to widespread false positives with standard MR techniques.
• The CAUSE method provides a more reliable approach for causal inference in the presence of complex pleiotropic effects.
• Shared heritable factors are prevalent and necessitate advanced methods like CAUSE for accurate causal discovery.