TEMR: Trans-ethnic Mendelian Randomization Method using Large-scale GWAS Summary Datasets

Abstract

AbstractAvailable large-scale GWAS summary datasets predominantly stem from European populations, while sample sizes for other ethnicities, notably Central/South Asian, East Asian, African, Hispanic, etc. remain comparatively limited, which induces the low precision of causal effect estimation within these ethnicities using Mendelian Randomization (MR). In this paper, we propose a Trans-ethnic MR method called TEMR to improve statistical power and estimation precision of MR in the target population using trans-ethnic large-scale GWAS summary datasets. TEMR incorporates trans-ethnic genetic correlation coefficients through a conditional likelihood-based inference framework, producing calibrated p-values with substantially improved MR power. In the simulation study, TEMR exhibited superior precision and statistical power in the causal effects estimation within the target populations than other existing MR methods. Finally, we applied TEMR to infer causal relationships from 17 blood biomarkers to four diseases (hypertension, ischemic stroke, type 2 diabetes and schizophrenia) in East Asian, African and Hispanic/Latino populations leveraging the biobank-scale GWAS summary data from European. We found that causal biomarkers were mostly validated by previous MR methods, and we also discovered 13 new causal relationships that were not identified using previously published MR methods.