In this work, we used a fully Bayesian polygenic modeling method, PRS-CSx, to derive SNP weights for a trans-ancestry PRS without the need of a priori population assignment or hyper-parameter tuning. PRS-CSx jointly models GWAS summary statistics across populations and explicitly accounts for population-specific allele frequencies and LD patterns. While non-European GWAS are often less powerful than European studies, they inform the genetic architecture in non-European populations and may capture population-specific genetic risk factors. Integrating available GWAS across ancestral groups may thus improve the portability of PRS -- especially to non-European populations -- and deliver personalized risk prediction that can more equally benefit all populations. Compared with early T2D PRS developed using a small number of SNPs selected based on statistical and/or biological significance (see [47] for a review), more recent T2D PRS derived from large-scale European GWAS [8, 9, 48] or multi-ethnic meta-GWAS [7, 44], and PRS evaluated in this work using more sophisticated PRS construction methods such as PRS-CS and LDpred2, our trans-ancestry PRS constructed by PRS-CSx demonstrated improved prediction accuracy and transferability across ancestral groups, reflecting the
