Key result
Transfer learning Mendelian randomization identifies ~93 shared causal metabolites for stroke across ancestries.
Why the study?
Most causal inferences from high-dimensional phenomes are based on European populations, limiting generalizability and equity across ancestries, and traditional MVMR methods struggle with high-dimensional correlated data.
Does a transfer learning framework for high-dimensional Mendelian randomization (TL-HDMR) improve the identification of shared and ancestry-specific causal metabolic biomarkers for stroke across diverse ancestral populations compared to traditional target-only HDMR methods?
Does a transfer learning framework for high-dimensional Mendelian randomization (TL-HDMR) improve the identification of shared and ancestry-specific causal metabolic biomarkers for stroke across diverse ancestral populations compared to traditional target-only HDMR methods?
Answers are grounded in this paper and cite their sources.
TL-HDMR provides a robust statistical framework that leverages cross-ancestry GWAS data to overcome small sample sizes and high dimensionality, enabling equitable discovery of shared and ethnic-specific causal metabolic biomarkers for stroke.
The limited genetic diversity in genome-wide association studies (GWAS) poses a significant challenge to the generalizability and equity of biomedical discoveries. Most causal inferences, particularly from high-dimensional phenomes (e.g. metabolomics), are primarily based on European populations, and their applicability to other ancestries remains uncertain. Traditional multivariable Mendelian randomization (MVMR) methods further struggle in high-dimensional and correlated settings due to collinearity and model instability. To bridge this gap, we present a two-step transfer learning framework for high-dimensional MR (TL-HDMR), designed to enhance causal exposure detection in understudied populations. Our approach leverages the Minimax Concave Penalty for asymptotically unbiased estimation amidst exposure correlations. Crucially, we introduce two novel pre-transfer procedures-HDMR.TSD for sourcing beneficial data and HDMR.PRESSO for filtering pleiotropic instruments-to ensure robust knowledge transfer. Extensive simulations demonstrated TL-HDMR's superior performance in ROC curves and mean absolute error over alternative methods. When applied to identify causal metabolites for stroke across multi-ancestry cohorts (European, East Asian, South Asian, and African), TL-HDMR successfully pinpointed both shared and ethnic-specific causal biomarkers, showcasing its unique capability for equitable causal inference. This work provides a powerful statistical tool that not only addresses critical methodological challenges but also promotes inclusivity and fairness in human health research.
No takes yet. Share an insight, caveat, or question.
Hou et al. (2026) studied Stroke and stroke subtypes (cardioembolic stroke, ischemic stroke, large artery stroke, and small vessel stroke). 899 metabolites evaluated via Two-step Transfer Learning High-Dimensional Mendelian Randomization (TL-HDMR) vs. Traditional target-only high-dimensional Mendelian randomization (HDMR) methods (LDA.Lasso, LDA.Elas, LDA.MCP) was evaluated on Identification of shared and ancestry-specific causal metabolic biomarkers for stroke, and simulation performance (ROC curve and mean absolute error [MAE]). TL-HDMR.MCP identified 93 shared causal metabolites for stroke across four ancestries alongside 37 European-, 34 African-, 25 South Asian-, and 17 East Asian-specific biomarkers.
Synapse has enriched 10 related papers. Consider them for comparative context: