NGS-based HLA genotyping
High-throughput genotyping of DNA samples was performed at the INSERM Unit UMR_S 1256 (NGERE, University of Lorraine, France) according to the manufacturer’s recommendations. We used the Illumina TruSight HLA v2 (Illumina Inc. CA, USA) sequencing panel, which provides high-resolution sequencing of 11 HLA loci (HLA-A , HLA-B , HLA-C ,HLA-DRB1/3/4/5 , HLA-DQB1 , HLA-DPB1 ,HLA-DQA1 , and HLA-DPA1 ), according to the manufacturer’s instructions at the INSERM Unit UMR_S 1256 (NGERE, University of Lorraine, France). Libraries were run in parallel on a MiSeq system (Illumina Inc. CA, USA) using 250 bp paired-end reads. We performed HLA allele assignation on the generated FASTQ files, using the TruSight HLA Assign 2.0 Software (Illumina Inc. CA, USA) to align the sequence reads and referencing with the International ImMunoGeneTics Information System/HLA database. We assessed the allele frequency of each HLA allele in the 24 patients studied. We calculated the enrichment ratio for a given HLA allele by dividing its allelic frequency found in the study population by that observed in the control population, using the Allele Frequency Net Database (www.allelefrequencies.net).19We defined the potentially enriched HLA alleles in patients with delayed hypersensitivity to penicillins as those with a frequency in the study population of at least 20% and an enrichment ratio >2. In the second phase of the study, we used an allelic model to estimate the odds ratio (OR) and the absolute risk difference for the association between carrying at least one DRB3*02:02 allele(DRB3*02:02:01:02 or DRB3*02:02:01:01 ) and the risk of delayed hypersensitivity to penicillins. We reported the 95% confidence interval (95% CI) and the associated P -values for the OR and the absolute risk difference using z-statistic and Chi-squared tests, respectively.