Abstract
The nomenclature of star alleles has been widely used in pharmacogenomics to enhance treatment outcomes, predict drug response variability, and reduce adverse reactions. However, the discovery of numerous rare functional variants through genome sequencing introduces complexities into the star-allele system. This study aimed to assess the nature and impact of the rapid discovery of numerous rare functional variants in the traditional haplotype-based star-allele system. We developed a new method to construct haplogroups, representing a common ancestry structure, by iteratively excluding rare and functional variants of the 25 representative pharmacogenes using the 2504 genomes from the 1000 Genomes Project. In total, 192 haplogroups and 288 star alleles were identified, with an average of 7.68 ± 4.2 cross-ethnic haplogroups per gene. Most of the haplogroups (70.8%, 136/192) were highly aligned with their corresponding classical star alleles (VI = 1.86 ± 0.78), exhibiting higher genetic diversity than the star alleles. Approximately 41.3% (N = 119) of the star alleles in the 2504 genomes did not belong to any of the haplogroups, and most of them (91.3%, 105/116) were determined by a single variant according to the allele-definition table provided by CPIC. These functional single variants had low allele frequency (MAF < 1%), high evolutionary conservation, and variant deleteriousness, which suggests significant negative selection. It is suggested that the traditional haplotype-based naming system for pharmacogenetic star alleles now needs to be adjusted by balancing both traditional haplotyping and newly emerging variant-sequencing approaches to reduce naming complexity.