as cases in the PRS analysis. After removing 1,776 REGARDS samples that overlapped with the MEDIA study, there was no remaining overlap between the discovery GWAS samples and participants in the evaluation cohorts. Figure 1 summarizes the construction and evaluation of the trans-ancestry T2D PRS derived by PRS-CSx in this work. Below, we briefly describe the phenotyping algorithm in eMERGE and the sample characteristics and processing of genetic data in each evaluation dataset.
