Mendelian randomization under the omnigenic architecture.

Area of Science:

• Genetics
• Epidemiology
• Statistical genetics

Background:

• Mendelian randomization (MR) is a key tool for causal inference in complex traits.
• Traditional MR methods rely on selecting a limited number of single nucleotide polymorphisms (SNPs) as instrumental variables.
• Complex traits are often influenced by thousands of SNPs, suggesting limitations in selecting only a few.

Purpose of the Study:

• To introduce a novel MR method, Omnigenic Mendelian Randomization (OMR), utilizing all genome-wide SNPs.
• To address the limitations of existing MR methods in capturing the polygenic or omnigenic architecture of complex traits.
• To provide a scalable computational framework for causal inference using all SNPs.

Main Methods:

• OMR employs summary statistics from genome-wide association studies (GWAS).
• The method incorporates a composite likelihood framework for efficient computation.
• It is designed to accommodate horizontal pleiotropy effects.

Main Results:

• Extensive simulations demonstrate OMR's power and robustness, even under model misspecification.
• Application to real data identified novel causal relationships for coronary artery disease (CAD) and asthma.
• Key factors identified include blood lipids, blood pressure, immunity for CAD, and immunity and obesity for asthma.

Conclusions:

• OMR offers a powerful new approach for causal inference in complex traits by leveraging all genome-wide SNPs.
• The method enhances our understanding of the genetic architecture underlying complex diseases like CAD and asthma.
• OMR facilitates the discovery of new etiological pathways and potential therapeutic targets.