Work

Building and validating a CYP2D6 reference panel for pharmacogenomic imputation

Year: 2026
Type: article
Abstract: CYP2D6 is a structurally complex pharmacogene responsible for metabolizing 20% of commonly prescribed drugs. In the UK Biobank (UKB), more than 10% of CYP2D6 alleles involve structural variations (SVs), including gene deletions and duplications that influence metabolizer status but are not reliably detectable using SNP arrays. This limitation excludes clinically important variation from large-scale pharmacogenomic studies. This thesis addresses this gap by developing a computational pipeline to generate fully phased CYP2D6 genotypes from UKB whole-genome sequencing (WGS) data and construct a large-scale reference panel (n = 480,354). The reference panel was evaluated by imputing CYP2D6 genotypes from SNP array positions in 1000 Genomes Project samples and validated using 122 GeT-RM samples with gold-standard consensus genotypes. For comparison, two pharmacogene genotype calling tools, Aldy 4 and PyPGx, were evaluated using WGS single nucleotide variants. The reference panel imputation accurately identified CYP2D6 genotypes in 94% (44/47) of samples of European ancestry. Imputation outperformed PyPGx (74%; 35/47) and Aldy (70%; 33/47), despite these tools having direct access to WGS variants. The improved performance was largely attributable to more accurate inference of SV-containing genotypes. Accuracy was lower in non-European samples (65%; 49/75), reflecting reference panel bias and the greater prevalence of complex SVs. Imputing CYP2D6 genotypes for 500,000 individuals was estimated to require only approximately 33 CPU hours, corresponding to a total computational cost of less than £1. The thesis also investigated associations between CYP2D6 metabolizer status, antidepressant switching, and the broader genetic architecture of switching phenotypes. Although no clear associations were observed, one genome-wide significant locus and several suggestive signals were identified, motivating future studies in larger or more precisely phenotyped cohorts. This thesis demonstrates that reference-panel imputation can accurately recover complex CYP2D6 genotypes, including clinically important SVs, at biobank scale. The approach provides a cost-effective framework for incorporating pharmacogenomic variation into existing array-based cohorts.
Language: English
Subfield: Pharmacology
Open Access status: green
OpenAlex ID: W7212408478
FWCI:
Cites:
Cited by:
Related to: 10