A multivariable cis-Mendelian randomization method robust to weak instrument bias and horizontal pleiotropy bias

Area of Science:

• Genetics and Bioinformatics
• Statistical Genetics
• Causal Inference

Background:

• Multivariable cis-Mendelian randomization (cis-MVMR) is valuable for identifying therapeutic targets influencing disease susceptibility.
• Existing cis-MVMR methods struggle with biases from weak instruments and horizontal pleiotropy, limiting their reliability.
• Invalid instruments can lead to inaccurate conclusions in genetic association studies.

Purpose of the Study:

• To develop a novel method, cis-Mendelian randomization bias correction estimating equation (cis-MRBEE), to overcome limitations in cis-MVMR.
• To improve the accuracy and robustness of identifying causal relationships between genetic variants and disease traits.
• To apply the new method to discover causal proteins for lipid traits and genetic factors for Alzheimer's disease.

Main Methods:

• Proposed cis-Mendelian randomization bias correction estimating equation (cis-MRBEE) leveraging local sparse genetic architecture.
• Employed fine-mapping to identify causal variants or exposure proxies.
• Utilized double-penalized minimization for estimating causal exposures and accounting for horizontal pleiotropy.

Main Results:

• Simulations demonstrated cis-MRBEE's robustness against weak instruments and horizontal pleiotropy, outperforming standard and existing cis-MVMR methods.
• Applied to the ANGPTL3 locus, cis-MRBEE identified APOA1, APOC1, and PCSK9 as likely causal proteins for LDL-C, HDL-C, and TG, revealing a complex lipid regulation network.
• Discovered CR1 expression in specific brain regions/cell types as potentially causal for Alzheimer's disease and its cerebrospinal fluid biomarkers (Aβ42, pTau).

Conclusions:

• cis-MRBEE offers a robust and reliable method for causal inference in genetic studies, effectively mitigating common biases.
• The study identified novel protein targets influencing lipid metabolism and potential genetic contributors to Alzheimer's disease pathogenesis.
• This approach advances the identification of therapeutic targets and understanding of complex disease mechanisms through genetic data analysis.