P077Imputing allelic variants for five human leukocyte antigen genes using a high-throughput genotyping array

Abstract

Aim To assess the feasibility of imputing human leukocyte antigen (HLA) alleles in African cohorts based on high-throughput genotyping of single-nucleotide polymorphisms (SNPs) in the human major histocompatibility complex (MHC). Methods For two African cohorts representing eastern and southern Africa, genomic DNA samples extracted from peripheral blood were first used for genotyping HLA-A , HLA-B , HLA-C , HLA-DRB1 and HLA-DQB1 alleles, all with 4-digit resolution through a combination of PCR-based genotyping techniques, including direct sequencing of locus-specific amplicons. SNP genotypes were further defined by a widely-used commercial array (the Illumina ImmunoChip). Paired HLA and SNP data from the first 400 individuals in each cohort served as a reference (the training set) for validation in the rest of samples. All imputation procedures were done using the HIBAG algorithm. Results A total of 7539 SNPs spanning the extended human MHC passed various quality control procedures and then merged with fully-resolved HLA alleles in 490 Rwandans and 1100 Zambians. Overall, SNP-based imputation of 2-digit allele groups at the five target loci showed success rates between 93.9% and 98.3%, while success rates in imputing 4-digit alleles ranged from 85.4% to 97.2%. Two major HLA-DRB1 allele groups (DRB1*04 and DRB1*10) in Zambians were the only ones with suboptimal imputation probabilities ( Conclusions In the era of high-throughput genomics, SNP-based imputation of HLA variants may offer a quick and affordable alternative for initial assessment of 2-digit allele groups. However, reliable imputation of 4-digit allele remains difficult even for some common alleles of broad interest.