Abstract
BACKGROUND: Accurate human leukocyte antigen (HLA) genotyping is a critical step in the implementation of neoantigen peptide-based cancer immunotherapy. Existing computational tools for HLA genotyping using high-throughput sequencing data often lack sufficient accuracy when used individually. Employing an ensemble of multiple software tools and sequencing sources can potentially address this limitation; however, this approach is hindered by increased processing time. METHODS: We evaluated an ensemble method that utilizes four HLA-genotyping software tools applied to three sequencing sources-tumor exome, normal exome, and tumor RNA-seq-from the same individual to achieve high-confidence HLA genotyping. To reduce processing time, we incorporated an HLA region-specific FASTQ read-filtering strategy. With this protocol we also provide a Docker implementation of the FASTQ read-filtering pipeline and a software tool for the analysis of HLA genotypes. RESULTS: Consensus HLA genotypes for HLA class I alleles, derived from filtered FASTQ files, showed complete concordance with those obtained from unfiltered original FASTQ files. The use of filtered FASTQ files significantly reduced the time required for HLA genotyping. CONCLUSIONS: These findings demonstrate the utility of the proposed HLA genotyping approach in achieving rapid and high-confidence two-field HLA genotyping.