Performance of a Modified ALLType NGS HLA Typing Assay with the Ion Chef/Ion S5/TypeStream Visual System

Abstract

Abstract OBJECTIVE S: The commercial ALLType assay (One Lambda, West Hills, CA) enables accurate, high-resolution typing of 11 HLA genes using the Ion Chef/Ion S5/TypeStream Visual (TSV) system. The assay features a multiplexed PCR to enrich all target HLA genes in a sample, and up to 48 samples can be indexed and sequenced in a run. The assay is laborious due to the late pooling of indexed libraries. Early pooling significantly simplifies the workflow but may increase the background noise and typing errors. This project aims to examine the feasibility of early pooling immediately after the indexing step and evaluate the performance of such a modified ALLType protocol. METHODS A total of 75 unique samples with diverse HLA genotypes were included in this study, which comprised 52 proficiency testing samples of three different sources, 10 stem cell recipients, 7 stem cell donors (including 2 buccal swabs), and 6 deceased organ donors. Genomic DNA extracted from these samples was tested in four sequencing runs, with 15-31 samples per run, using the modified ALLType procedure (reagent lot: 15). The results were compared with the reference typings generated by the manufacturer-established protocol. All data were analyzed using TSV (version 3.0) and the IPD-IMGT/HLA library (version 3.47); the default analysis parameters were applied. The primary endpoint of the study was the concordance of 3-field typing results. The secondary endpoint was key health metrics, including allele balance, key-exon coverage, and background noise. We also evaluated the saving on hands-on time and reagents with the modified protocol. RESULTS The concordance rates of 3-field typing results were 100% for HLA-A, HLA-B, HLA-C, HLA-DPA1, HLA-DPB1, HLA-DQA1, HLA-DQB1, and HLA-DRB1, and 96.2% for HLA-DRB3/4/5. The three discrepant HLA-DRB3/4/5 results with the modified protocol were false positive allele calls due to a small number of contaminating reads, which were corrected by increasing the threshold for HLA-DRB3/4/5 in TSV. The allele balance and exon 2 coverage across all samples and genes were comparable between the modified and reference protocols, with R-squared values of 0.83 and 0.81, respectively. High background signals in exons were observed in only 0.6% of all samples with both protocols. The modified protocol reduced the hands-on time by 15%, from 6.5 to 5.5 hours, and reduced the use of library preparation reagents by 30%. CONCLUSION It is feasible to pool indexed libraries early during the library preparation process without increasing the background noise in the sequencing data. The modified ALLType protocol showed typing accuracy and health metrics comparable to the original protocol, with significant savings on time and reagents.