Streaming Long-Read Sequence Alignments for HLA Predictions Using HLAminer

Abstract

Long-read sequencing platforms such as the Oxford Nanopore Technologies (ONT) and Pacific Biosciences (PacBio) platforms now offer sufficient read lengths, throughput, and accuracy at competitive costs to analyze polymorphic regions of the human genome, including the highly complex human leukocyte antigen (HLA) gene cluster—a cornerstone of human immunity. Here, we present a streamlined protocol for predicting HLA signatures from whole-genome shotgun (WGS) long-read sequencing data by directly streaming sequence alignments into HLAminer. This method is as simple as running minimap2, scales efficiently with the number of sequences, and works with any read aligner compatible with the SAM file format—eliminating the need to store bulky alignment files on disk. We provide a step-by-step guide for predicting HLA class I and class II alleles from third-generation long-read sequencing data and demonstrate the robustness of predictions even with older, less accurate WGS nanopore datasets and relatively low (10×) sequencing coverage. Code availability: HLAminer is available under the BC Cancer software license agreement (academic use) at https://github.com/bcgsc/HLAminer. © 2025 The Author(s). Current Protocols published by Wiley Periodicals LLC.

Basic Protocol: HLA prediction from streamed ONT or PacBio long-read alignments