300 clinical samples from five labs were sequenced at five loci (HLA-A, B, C, DRB1 and DQB1) for a total of 3000 alleles using Holotype HLA on an Illumina MiSeq using the 2x250 bp chemistry. The resulting sequencing data was analyzed and the HLA allele calls were assigned using Omixon’s HLA Twin software. The reference typing results had been previously obtained by SBT or a combination of SBT/SSO/SSP. The NGS results from Holotype HLA were compared to the existing typings to determine concordance among the methods.
Six alleles had discordant HLA calls between the NGS result and the SBT ‘reference’ typing. Five of these alleles were discordant at the second field and one sample was discordant at the third field. Mistypings were observed in HLA-A, HLA-C, HLA-DQB1 and HLA-DRB1 among the six discordant samples. The errors were due to a variety of reasons that ranged from failure of SBT to detect a second allele in a sample to noisy SBT trace. Retyping these by SBT confirmed that the NGS result provided the correct typing.
NGS is a suitable method for high resolution HLA genotyping. During this comparison we uncovered a number of incorrect SBT typings which NGS can reliably genotype without ambiguity, thus highlighting the importance of adopting this new technique in the clinic. In addition, we were able to fill in previously unknown regions of each allele by generating whole gene consensus sequences.