A comprehensive assessment of pharmacogenomic annotation tools for next-generation sequencing data: an emphasis on...

Next-generation sequencing (NGS) offers a fast, cost-effective, and scalable solution for pharmacogenomic allele assignment but faces challenges in accurately identifying variants and haplotypes in regions with high sequence similarity. This study aimed to evaluate the performance of three NGS-compatible genotyping tools - PyPGx, Stargazer, and Aldy - for CYP2D6 annotation and investigated the CYP2D6 genetic distribution in 1008 whole-genome sequences from the 1000 Vietnamese Genome Project (VN1K) data. A benchmark dataset was constructed using 8556 diverse CYP2D6 alleles and 122 samples with complex structural variations. Tools were then assessed for haplotype, diplotype, and phenotype concordance across sequencing coverages of 8x, 30x, and 60x. Then, all three tools were subsequently applied to genotype CYP2D6 in the VN1K dataset. Overall, Aldy outperformed others, achieving haplotype and phenotype accuracies of 89.56% and 96.59%, respectively, even at low coverage (8x). In comparison, Stargazer and PyPGx achieved diplotype concordance rates of 51.33% and 47.77% under the same coverage level. According to the output of Aldy - the best performance tool in analysing the CYP2D6 genetic distribution in VN1K data, the results revealed a predominance of reduced-function alleles and a high prevalence of intermediate metabolizer phenotypes, underscoring the need for population-specific pharmacogenomic strategies and highlighting Aldy's potential in advancing precision medicine.