Generation of high confidence HLA genotyping and consensus sequences for class I HLA loci using the NGS-based Omixon Holotype HLA typing system and the Illumina Miseq platform

Abstract

Aim Generate HLA genotyping and high confidence consensus sequences for 50 samples at HLA-A, B and C loci using the Omixon Holotype HLA typing system. Methods The Omixon Holotype HLA typing kit was used to generate sequencing libraries for 50 samples for HLA-A, B and C loci, and sequenced on an Illumina MiSeq. HLA typing and consensus sequence generation was performed using Omixon Twin, GenDx NGSengine and a custom pipeline, which relies on stringently mapping reads to the genotyped alleles and using SAMtools to generate the consensus. Consensus sequences generated by each method were aligned pairwise to determine concordance using a semi-global Needleman–Wunsch algorithm. When the sequences generated by each method were concordant, sequences were aligned to the IMGT reference sequence of the genotyped allele, to annotate novel alleles and annotate the sequence of incomplete alleles. When sequences generated by each method were discordant, sequences were aligned to the genotyped allele for manual inspection and assessment of algorithm performance. Results HLA genotyping results for all 300 alleles (50 samples at 3 loci) were 100% concordant between NGSengine and Omixon Twin, after manually resolving discrepancies (6/300). Consensus sequences were generated for every allele (300) encompassing 31 unique HLA-A, 52 unique HLA-B and 31 unique HLA-C alleles. Our analysis facilitated the annotation of full gene sequences for 11 HLA alleles for which there is no complete IMGT reference sequence (A∗02:04, B∗14:03, B∗15:08:01, B∗15:30, B∗27:02:01, B∗35:04:01, B∗35:43:01, B∗39:09:01, B∗51:08:01, B∗58:02 and C∗07:18) and facilitated the discovery of two novel HLA-A alleles, with intronic substitutions. For the remaining alleles, consensus sequences were compared against the reference sequences of the genotyped alleles and occasional differences between consensus sequences and reference sequences were manually inspected to evaluate algorithm performance. It has been determined that depth of coverage is critical in phasing and base calling, which is necessary to generate credible consensus sequences. Conclusions Generation of high confidence HLA class I consensus sequences is possible using the Omixon Holotype HLA typing system provided that adequate depth of uniform coverage is obtained from sequencing. D. Monos: Other (Identify); Company/Organization; Omixon Royalties.