PGx-Synthea: a population-calibrated synthetic EHR–pharmacogenomic cohort and PGx-CDS Bench v1.1

Fully synthetic linked EHR–pharmacogenomic cohort and benchmark, generated by PGx-Synthea. The canonical run contains 11,494 synthetic patients with ancestry-stratified diplotypes for seven CPIC Tier-1 genes, CPIC phenotype translation, and RxNorm-linked drug–gene discordance alerts, together with a sample of HL7 FHIR R4 Genomics Reporting bundles and PGx-CDS Bench v1.1 (148 cases with an expected-output oracle, shipped with one FHIR Genomics bundle per case). No real patient data were used; the generator reads only published aggregate allele frequencies. Generated data and reference tables are released under CC BY-SA 4.0, matching the ClinPGx source licence; the code that produced them is MIT-licensed at the linked repository.