🎙️ Episode 18: Network Causality — A New Bayesian Model for Multivariable Mendelian Randomization

🎙️ Episode 18: Network Causality — A New Bayesian Model for Multivariable Mendelian Randomization

🧬 In this episode of Base by Base, we examine the methodological breakthrough presented by Liu et al. (2025) in Nature Communications. The authors introduce BayesMRnet, a Bayesian causal graphical model designed to perform joint Mendelian randomization (MR) across multiple exposures and outcomes. By leveraging genome-wide association study (GWAS) data, BayesMRnet constructs directed acyclic graphs (DAGs) that capture both direct and mediated causal effects among phenotypic traits, even in the presence of horizontal pleiotropy.

🔍 Key highlights: BayesMRnet departs from traditional pairwise MR by modeling causal relationships among multiple phenotypes simultaneously, using genetic variants as instrumental variables; it employs a DAG framework combined with Bayesian inference to disentangle direct and indirect effects, accommodating pleiotropic and mediated pathways; extensive simulations demonstrate that BayesMRnet achieves higher sensitivity and lower false-positive rates compared with established methods such as MR-Egger and multivariable MR; applied to real-world data on 66 metabolic and anthropometric traits, the model reconstructed biologically plausible causal networks, revealing novel pathways influencing cardiometabolic risk; and its Bayesian foundation allows quantification of uncertainty on each network edge, offering more robust interpretation in complex clinical and epidemiological settings.

🧠 Conclusion: BayesMRnet ushers in a new era for Mendelian randomization by enabling multivariate modeling of exposures and outcomes at scale. By integrating structural dependencies and uncertainty estimates, this framework provides a powerful tool for uncovering causal relationships in human genomics and guiding more effective prevention strategies.

📖 Reference: Liu, Y., Xiang, Z., Han, Z., et al. (2025). Bayesian causal graphical model for joint Mendelian randomization analysis of multiple exposures and outcomes. Nature Communications, 16, 3142. https://doi.org/10.1038/s41467-024-46789-7

📜 License: This episode is based on an open access article published under the Creative Commons Attribution 4.0 International License (CC BY 4.0).