P207 Evaluation of Oxford Nanopore MinION sequencing to predict antimicrobial resistance profiles in clinical N. gonorrhoeae strains

Abstract

<h3>Background</h3> Antimicrobial resistance (AMR) in N. gonorrhoeae (NG) is a big concern for public health and has an impact on treatment efficiency. Whole genome sequencing (WGS) of clinical strains can be used to predict AMR profiles based on detection of known resistance-associated mutations (RAM) and AMR gene acquisition. While Illumina short read sequencing gives low error rates, third-generation sequencing with Oxford Nanopore Technologies (ONT) has the advantage of generating long reads in real-time. We aimed to evaluate the ability of ONT sequencing to predict AMR in NG. <h3>Methods</h3> WGS was performed on one azithromycin resistant and four ceftriaxone resistant clinical NG strains using ONT (MinION R9.4 flowcell) and Illumina (MiSEQ) sequencing platforms. Assemblies were obtained from: Illumina data (by SPAdes), ONT data (by Flye) and a hybrid assembly of both ONT and Illumina data (by Unicycler). Moreover, we evaluated four different polishing strategies (based on rebaler, racon and/or medaka) after Flye. Completeness of the assemblies were visually checked with bandage. AMR profile prediction based on RAM detection and AMR gene acquisition were obtained by using pathogenwatch and compared with phenotypic antimicrobial susceptibility testing. <h3>Results</h3> Hybrid and ONT assemblies showed a complete, closed chromosome, while Illumina assemblies were not able to resolve genome structures. Pathogenwatch could predict AMR profiles (except for CRO) correctly on Illumina and hybrid assemblies, and detected RAM were concordant in both assemblies. RAM detection on ONT assemblies was inconsistent among the different polishing strategies. Moreover, fewer RAMs were detected which resulted in missed resistance in the predicted AMR profiles. <h3>Conclusion</h3> Here, we demonstrate that ONT sequencing reveals the genomic structure of NG isolates, but high error rates leads to it missing RAMs in the consensus assembly. Polishing strategies of MinION based assemblies can reduce the proportion of RAMs missed but remain inferior to Illumina based assemblies.