Abstract
Pharmacogenomics enables the personalization of drug therapy by linking genetic variations to differences in drug metabolism, efficacy, and risk of adverse reactions. Genetic polymorphisms within cytochrome P450 (CYP) genes significantly affect enzyme activity, influencing drug plasma levels, responses, and safety. Central to this process is accurate genotype-to-phenotype translation, especially for the CYP enzyme family, which metabolizes 70-80% of clinically used drugs. To address this, we have developed GenoStaR, an R package that converts genotypes into star alleles and predicts the associated metabolizer status for major cytochrome P450 genes-CYP1A2, CYP2B6, CYP2C9, CYP2C19, CYP2D6, CYP3A4, and CYP3A5. GenoStaR assigns star alleles using single-nucleotide polymorphisms, insertion-deletion variants, and structural variants. Given genotype data, GenoStaR uses comprehensive allele definition tables to determine diplotypes, activity scores, and predicted metabolizer status. The tool accounts for complex scenarios, including CYP2D6 copy number variations, using a tiered matching strategy and structural variant detection. We evaluated GenoStaR using two datasets. The first from the Centre for Addiction and Mental Health Individualized Medicine: Pharmacogenetics Assessment and Clinical Treatment (IMPACT) study (n = 8,287), which included genotyping data, along with star allele information from a commercial pharmacogenetic test. The second, the Toronto Schizophrenia sample (n = 188), with in-house genotype data and manually validated star alleles. GenoStaR achieved 100% concordance in diplotype calls across both datasets. GenoStaR offers a reliable, efficient, and accurate solution for converting genotypes into star alleles and predicting CYP-related metabolizer status. Its performance on a large validation dataset highlights its potential to enhance pharmacogenomic testing in clinical settings.