Construction and benchmarking of a multi-ethnic reference panel for the imputation of HLA class I and II alleles

Abstract

Genotypeimputationof the human leukocyte antigen (HLA) region is a cost-effective means to infer classicalHLAalleles from inexpensive and dense SNP array data. In the research setting,imputationhelps avoid costs for wet lab-basedHLAtyping and thus renders association analyses of theHLAin large cohorts feasible. Yet, mostHLAimputationreferencepanels target Caucasian ethnicities andmulti-ethnicpanels are scarce. We compiled a high-qualitymulti-ethnicreferencepanelbased on genotypes measured with Illumina's Immunochip genotyping array andHLAtypes established using a high-resolution next generation sequencing approach. Ourreferencepanelincludes more than 1,300 samples from Germany, Malta, China, India, Iran, Japan and Korea and samples of African American ancestry for all classicalHLAclassI and II alleles includingHLA-DRB3/4/5. Applying extensive cross-validation, we benchmarked theimputationusing theHLAimputationtool HIBAG, ourmulti-ethnicreferenceand an independent, previously published data set compiled of subpopulations of the 1000 Genomes project. We achieved averageimputationaccuracies higher than 0.924 for the commonly studiedHLA-A, -B, -C, -DQB1 and -DRB1 genes across all ethnicities. We investigated allele-specificimputationchallenges in regard to geographic origin of the samples using sensitivity and specificity measurements as well as allele frequencies and identifiedHLAalleles that are challenging to impute for each of the populations separately. In conclusion, our newmulti-ethnicreferencedata set allows for high resolutionHLAimputationof genotypes at all classicalHLAclassI and II genes including theHLA-DRB3/4/5 loci based on diverse ancestry populations.