Towards allele-level human leucocyte antigens genotyping - assessing two next-generation sequencing platforms: Ion...

Human leucocyte antigens (HLA) typing has been a challenge due to extreme polymorphism of the HLA genes and limitations of the current technologies and protocols used for their characterization. Recently, next-generation sequencing techniques have been shown to be a well-suited technology for the complete characterization of the HLA genes. However, a comprehensive assessment of the different platforms for HLA typing, describing the limitations and advantages of each of them, has not been presented. We have compared the Ion Torrent Personal Genome Machine (PGM) and Illumina MiSeq, currently the two most frequently used platforms for diagnostic applications, for a number of metrics including total output, quality score per position across the reads and error rates after alignment which can all affect the accuracy of HLA genotyping. For this purpose, we have used one homozygous and three heterozygous well-characterized samples, at HLA-A, HLA-B, HLA-C, HLA-DRB1 and HLA-DQB1. The total output of bases produced by the MiSeq was higher, and they have higher quality scores and a lower overall error rate than the PGM. The MiSeq also has a higher fidelity when sequencing through homopolymer regions up to 9 bp in length. The need to set phase between distant polymorphic sites was more readily achieved with MiSeq using paired-end sequencing of fragments that are longer than those obtained with PGM. Additionally, we have assessed the workflows of the different platforms for complexity of sample preparation, sequencer operation and turnaround time. The effects of data quality and quantity can impact the genotyping results; having an adequate amount of good quality data to analyse will be imperative for confident HLA genotyping. The overall turnaround time can be very comparable between the two platforms; however, the complexity of sample preparation is higher with PGM, while the actual sequencing time is longer with MiSeq.