Mendelian randomization while jointly modeling cis genetics identifies causal relationships between gene expression...

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Mendelian randomization (MR) infers causality between gene expression and complex traits.
• Pleiotropy and linkage disequilibrium (LD) of gene-expression quantitative trait loci (eQTL) confound MR analyses.
• Existing MR methods struggle with unobserved pleiotropy and LD, especially with limited eQTL data.

Purpose of the Study:

• To propose MR-link, a novel MR method designed to account for unobserved pleiotropy and LD.
• To leverage individual-level data for robust causal inference in transcriptome-wide association studies.
• To improve the accuracy and power of MR analyses for complex traits.

Main Methods:

• MR-link utilizes individual-level data to address confounding factors like pleiotropy and LD.
• The method is effective even when only a single eQTL variant is available.
• Performance was evaluated through simulations and compared against established MR techniques and coloc.

Main Results:

• Simulations demonstrated MR-link's low false-positive rates (median 0.05) and high statistical power (up to 0.89).
• MR-link outperformed all other tested MR methods and coloc in simulation studies.
• Application to low-density lipoprotein cholesterol (LDL-C) in 12,449 individuals identified 25 genes causally linked to LDL-C.

Conclusions:

• MR-link offers a robust approach for transcriptome-wide causal inference.
• The method successfully identified known and novel genes associated with LDL-C, including PVRL2 in the APOE locus.
• MR-link enhances the ability to uncover causal relationships between gene expression and complex traits.