Mendelian Randomization & Causal Inference
Mendelian Randomization
Using large-scale cohort GWAS summary statistics, the Mendelian randomization framework assesses causal relationships between exposures and outcomes. Integrates public databases such as UK Biobank and FinnGen, automating instrumental variable selection, causal effect estimation, and multi-dimensional sensitivity testing.
Applicable Scenarios
- Drug target causal validation
- Biomarker causal direction determination
- Disease risk factor screening
- Protein-disease causal association
Data Inputs
- Exposure GWAS summary statistics
- Outcome GWAS summary statistics
- Reference panel (e.g., 1000 Genomes)
Deliverables
- Causal effect estimates (IVW, MR-Egger, Weighted Median)
- Leave-one-out sensitivity analysis forest plot
- Funnel plot and heterogeneity test
- Colocalization posterior probability
Analysis Pipeline
IV Selection
Genome-wide significant SNP screening, LD pruning (r² < 0.001)
Harmonization
Allele direction unification, palindromic SNP handling
MR Analysis
IVW, MR-Egger, Weighted Median, MR-PRESSO
Sensitivity
Leave-one-out, funnel plot, Cochran's Q heterogeneity test
Colocalization
coloc posterior probability analysis
Reporting
Methodology description, results table, visualization charts
Related Case
Biomarker Screening & Diagnostic Model
Integrated large-scale cohort data from UKB-PPP, FinnGen, and built an automated causal inference pipeline for candidate protein screening and disease correlation assessment.
Deliverables
- MR Analysis Pipeline
- Colocalization & Sensitivity Tests
- Candidate Protein Ranking Table
- Methodological QC Records