Easy-HLA: a validated web application suite to reveal the full details of HLA typing

Abstract

Motivation: The HLA system plays a pivotal role in both clinical applications and immunology research. Typing HLA genes in patient and donor is indeed required in hematopoietic stem cell and solid-organ transplantation, and the histocompatibility complex region exhibits countless genetic associations with immune-related pathologies. Since the discovery of HLA antigens, the HLA system nomenclature and typing methods have constantly evolved, which leads to difficulties in using data generated with older methodologies.

Results: Here, we present Easy-HLA, a web-based software suite designed to facilitate analysis and gain knowledge from HLA typing, regardless of nomenclature or typing method. Easy-HLA implements a computational and statistical method of HLA haplotypes inference based on published reference populations containing over 600 000 haplotypes to upgrade missing or partial HLA information: 'HLA-Upgrade' tool infers high-resolution HLA typing and 'HLA-2-Haplo' imputes haplotype pairs and provides additional functional annotations (e.g. amino acids and KIR ligands). We validated both tools using two independent cohorts (total n = 2500). For HLA-Upgrade, we reached a prediction accuracy of 92% from low- to high-resolution of European genotypes. We observed a 96% call rate and 76% accuracy with HLA-2-Haplo European haplotype pairs prediction. In conclusion, Easy-HLA tools facilitate large-scale immunogenetic analysis and promotes the multi-faceted HLA expertise beyond allelic associations by providing new functional immunogenomics parameters.

Availability and implementation: Easy-HLA is a web application freely available (free account) at: https://hla.univ-nantes.fr.

Supplementary information: Supplementary data are available at Bioinformatics online.

Fig. 1.

Easy-HLA software presentation. (A) Example of the single query mode. The patient genotype is entered for each gene in first or second-field, serology, NMDP codes or left empty. The user must choose the output resolution (2: second-field, 1: first-field, 0: empty), the post-probability threshold and reference population. The output table contains full mid to high-resolution genotypes with their respective likelihood and post-probability. (B) Easy-HLA delivers updated HLA information from low-resolution HLA typing. In this example, we start with a classical HLA serological genotype (A∼B∼DRB1∼DQB1). HLA-Upgrade statistically predicts high-resolution genotypes (left panel), and can also predict an untyped locus, such as HLA-C (middle panel). Finally, HLA-2-Haplo imputes the most likely haplotype pair. These predictions are all done in silico and as such prevent from additional genotyping in the laboratory

Fig. 2.

Validation of the HLA-Upgrade module in the European (EUR, left) and African-American (AFA, right) populations (post-probability threshold set at 0%). HLA-A∼B∼C∼DRB1∼DQB1 high-resolution genotypes were predicted from different gene combinations of first-field genotypes: HLA-A∼B∼DRB1 (blue), HLA-A∼B∼C∼DRB1 (red), HLA-A∼B∼C∼DRB1∼DQB1 (dark blue). (A) Prediction accuracy per locus. (B) Prediction accuracy according to genotype post-probability. (C) Call rate according to genotype post-probability. (D) Prediction accuracy according to call rate. (Color version of this figure is available at Bioinformatics online.)

Fig. 3.

Validation of the HLA-2-Haplo module in the European (EUR, dark blue) and African-American (AFA, red) populations. (A) Accuracy of haplotypes pairs prediction according to the calculated post-probability. (B) Call rate of haplotypes pairs prediction according to the calculated post-probability. (C) Accuracy of haplotypes pairs prediction according to the call rate. (Color version of this figure is available at Bioinformatics online.)