MR-link2

MR-link2 estimates causal effects and pleiotropy in Mendelian randomization analyses using summary statistics and linkage disequilibrium (LD) information.

Key Features:

• Simultaneous estimation: Estimates causal effect and pleiotropy in a single associated region.
• Calibrated Type 1 error: Has calibrated Type 1 error rates and high power in realistic simulations.
• Superior performance metrics: Achieves superior area under the receiver operator characteristic curve (AUC) up to 0.80 in metabolic pathway replication and AUC 0.90 in causal direction testing between blood cell type compositions and gene expression.
• Reduced heterogeneity: Produces several fold less heterogeneous causal effect estimates compared to competing methods.
• Summary statistics input: Operates on summary statistics and LD information without requiring individual-level data.

Scientific Applications:

• Metabolic pathway analysis: Replicates causal relationships derived from metabolic pathway references using metabolite quantitative trait locus studies.
• Complex trait causality: Tests canonical causal relationships between complex traits with lower per-locus Type 1 error rates.
• Gene expression and cell composition: Determines causal direction between blood cell type compositions and gene expression of their marker genes.
• Novel metabolite discovery: Identifies causal links between metabolites not connected by canonical reactions, such as between glycine and pyrroline-5-carboxylate.

Methodology:

MR-link2 uses summary statistics and linkage disequilibrium (LD) information to simultaneously estimate a causal effect and pleiotropy in a single associated region.