9: MrDAG and the causal architecture of mental health

Zuber V et al., The American Journal of Human Genetics - Zuber et al. introduce MrDAG, a Bayesian causal graphical model that combines Mendelian randomization, structure learning, and interventional calculus to estimate causal effects among multiple correlated exposures and outcomes using summary-level GWAS data. The method reveals dependency structures and highlights education and smoking as key intervention points for mental health. Key terms: Mendelian randomization, Bayesian networks, causal inference, mental health, genetics.

Study Highlights:
MrDAG learns unconfounded dependency relations within exposures and outcomes by using genetically predicted trait components and explores essential graphs under the constraint that exposures causally precede outcomes. The model integrates structure learning with MR instrumental-variable logic and estimates causal effects via interventional calculus, averaging over graph uncertainty through Bayesian model averaging. In simulations MrDAG outperformed one-outcome-at-a-time and other multivariate MR and graphical approaches, showing fewer false positives and lower bias in causal effect estimates. Applied to lifestyle exposures and mental health phenotypes, MrDAG identified education and smoking as primary actionable nodes and uncovered mediated paths linking smoking to schizophrenia liability and cognition.

Conclusion:
MrDAG provides a scalable Bayesian framework to map complex causal pathways among multiple exposures and outcomes from summary genetic data, improving direct causal effect estimation and prioritizing interventions such as education and smoking reduction for mental health.

QC:
This episode was checked against the original article PDF and publication metadata for the episode release published on 2025-04-18.

QC Scope:
- article metadata and core scientific claims from the narration
- excludes analogies, intro/outro, and music

QC Summary:
- factual score: 10/10
- metadata score: 10/10
- supported core claims: 7
- claims flagged for review: 0
- metadata checks passed: 4
- metadata issues found: 0

Metadata Audited:
- article_doi
- article_title
- article_journal
- license

Factual Items Audited:
- MrDAG is a Bayesian causal graphical model for joint Mendelian randomization analysis of multiple exposures and outcomes
- Directionality from exposures to outcomes is assumed with no reverse causation
- Six lifestyle exposures and seven mental health outcomes are modeled
- Education and smoking are primary intervention points
- A causal path exists from smoking to MDD to BD to schizophrenia
- Ascertainment bias explains the education-ASD association

QC result: Pass.