Characterisation of pharmacogene allelic variation in African populations and development of a novel diplotype calling algorithm

Abstract

Genetic variation is in part responsible for the variability in drug response within and between populations. Therefore, there is a shift in drug therapy in the form of implementing genotypeguided treatment in place of the ‘one-size-fits-all’ model so as to promote drug efficacy and reduce the risk of adverse drug reactions. However, the full catalogue of pharmacogenetic variants is yet to be established, in particular for African populations. Therefore, this study aimed to characterise the variation in three core pharmacogenes in drug metabolism i.e. CYP2D6, CYP2B6 and CYP2A6, across diverse continental African populations.

Given the known challenges of genotyping these hypervariable genes, a benchmark of the stateof-the-art star allele calling bioinformatics algorithms was performed using simulated and realworld whole genome sequence (WGS) data. Furthermore, we developed a novel graph- and Nextflow-based pipeline (StellarPGx) to facilitate scalable, reproducible and more accurate CYP genotyping using WGS data. Thereafter, we applied a consensus star allele calling approach, involving StellarPGx and the other existing tools, to assess the CYP2D6, CYP2B6 and CYP2A6 allelic diversity mainly across sub-Saharan Africa, based on 962 high-depth African genomes. Targeted Single Molecule Real-Time (SMRT) Sequencing was used to validate novel CYP2D6, CYP2B6 and CYP2A6 haplotypes found in individuals whose DNA samples were available to us through various collaborations.

Our analysis highlighted the frequency of known normal, decreased, increased and no-function CYP2D6, CYP2B6 and CYP2A6 alleles across sub-Saharan Africa in comparison with other global populations. Differences in predicted CYP2D6 and CYP2B6 metaboliser phenotypes were observed even between some populations from neighbouring geographical regions. In addition, from the short-read WGS data from African individuals, we inferred over 77 potential novel African-specific alleles in CYP2D6, CYP2B6 and CYP2A6 combined. Targeted SMRT sequencing for a subset of the study population enabled further characterisation of two novel African-specific alleles apiece for CYP2D6, CYP2B6 and CYP2A6.

This study highlights the importance of combining multiple tools in accurately mining high coverage WGS data for variation in highly polymorphic CYP genes. Furthermore, our novel bioinformatics tool, StellarPGx, addresses the need to implement a pipeline for accurate, scalable, and reproducible pharmacogene allele calling using WGS data. The variability in pharmacogene allele and phenotype distributions across the African populations represented in this study, and the identification of several novel alleles underscore the need for investigating pharmacogene variation in an African context to reliably inform clinical pharmacogenomics implementation strategies across Africa.