Mycobacterium tuberculosis resistance to anti-tuberculosis drugs is a major threat to global public health. Whole genome sequencing (WGS) is rapidly gaining traction as a diagnostic tool for clinical tuberculosis settings. To support this informatically, previous work led to the development of the widely used TBProfiler webtool, which predicts resistance to 14 drugs from WGS data. However, for accurate and rapid high throughput of samples in clinical or epidemiological settings, there is a need for a stand-alone tool and the ability to analyse data across multiple WGS platforms, including Oxford Nanopore MinION. We present a new command line version of the TBProfiler webserver, which includes hetero-resistance calling and will facilitate the batch processing of samples. The TBProfiler database has been expanded to incorporate 178 new markers across 16 anti-tuberculosis drugs. The predictive performance of the mutation library has been assessed using > 17,000 clinical isolates with WGS and laboratory-based drug susceptibility testing (DST) data. An integrated MinION analysis pipeline was assessed by performing WGS on 34 replicates across 3 multi-drug resistant isolates with known resistance mutations. TBProfiler accuracy varied by individual drug. Assuming DST as the gold standard, sensitivities for detecting multi-drug-resistant TB (MDR-TB) and extensively drug-resistant TB (XDR-TB) were 94% (95%CI 93–95%) and 83% (95%CI 79–87%) with specificities of 98% (95%CI 98–99%) and 96% (95%CI 95–97%) respectively. Using MinION data, only one resistance mutation was missed by TBProfiler, involving an insertion in the tlyA gene coding for capreomycin resistance. When compared to alternative platforms (e.g. Mykrobe predictor TB, the CRyPTIC library), TBProfiler demonstrated superior predictive performance across first- and second-line drugs. The new version of TBProfiler can rapidly and accurately predict anti-TB drug resistance profiles across large numbers of samples with WGS data. The computing architecture allows for the ability to modify the core bioinformatic pipelines and outputs, including the analysis of WGS data sourced from portable technologies. TBProfiler has the potential to be integrated into the point of care and WGS diagnostic environments, including in resource-poor settings.

The emergence of resistance to anti-tuberculosis drugs is a serious and growing threat to public health. Next-generation sequencing is rapidly gaining traction as a diagnostic tool for investigating drug resistance in Mycobacterium tuberculosis to aid treatment decisions. However, there are few little data regarding the precision of such sequencing for assigning resistance profiles. We investigated two sequencing platforms (Illumina MiSeq, Ion Torrent PGM™) and two rapid analytic pipelines (TBProfiler, Mykrobe predictor) using a well characterised reference strain (H37Rv) and clinical isolates from patients with tuberculosis resistant to up to 13 drugs. Results were compared to phenotypic drug susceptibility testing. To assess analytical robustness individual DNA samples were subjected to repeated sequencing. The MiSeq and Ion PGM systems accurately predicted drug-resistance profiles and there was high reproducibility between biological and technical sample replicates. Estimated variant error rates were low (MiSeq 1 per 77 kbp, Ion PGM 1 per 41 kbp) and genomic coverage high (MiSeq 51-fold, Ion PGM 53-fold). MiSeq provided superior coverage in GC-rich regions, which translated into incremental detection of putative genotypic drug-specific resistance, including for resistance to para-aminosalicylic acid and pyrazinamide. The TBProfiler bioinformatics pipeline was concordant with reported phenotypic susceptibility for all drugs tested except pyrazinamide and para-aminosalicylic acid, with an overall concordance of 95.3%. When using the Mykrobe predictor concordance with phenotypic testing was 73.6%. We have demonstrated high comparative reproducibility of two sequencing platforms, and high predictive ability of the TBProfiler mutation library and analytical pipeline, when profiling resistance to first- and second-line anti-tuberculosis drugs. However, platform-specific variability in coverage of some genome regions may have implications for predicting resistance to specific drugs. These findings may have implications for future clinical practice and thus deserve further scrutiny, set within larger studies and using updated mutation libraries.

Despite the advent of whole genome metagenomics, targeted approaches (such as 16S rRNA gene amplicon sequencing) continue to be valuable for determining the microbial composition of samples. Amplicon microbiome sequencing can be performed on clinical samples from a normally sterile site to determine the aetiology of an infection (usually single pathogen identification) or samples from more complex niches such as human mucosa or environmental samples where multiple microorganisms need to be identified. The methodologies are frequently applied to determine both presence of micro-organisms and their quantity or relative abundance. There are a number of technical steps required to perform microbial community profiling, many of which may have appreciable precision and bias that impacts final results. In order for these methods to be applied with the greatest accuracy, comparative studies across different laboratories are warranted. In this study we explored the impact of the bioinformatic approaches taken in different laboratories on microbiome assessment using 16S rRNA gene amplicon sequencing results. Data were generated from two mock microbial community samples which were amplified using primer sets spanning five different variable regions of 16S rRNA genes. The PCR-sequencing analysis included three technical repeats of the process to determine the repeatability of their methods. Thirteen laboratories participated in the study, and each analysed the same FASTQ files using their choice of pipeline. This study captured the methods used and the resulting sequence annotation and relative abundance output from bioinformatic analyses. Results were compared to digital PCR assessment of the absolute abundance of each target representing each organism in the mock microbial community samples and also to analyses of shotgun metagenome sequence data. This ring trial demonstrates that the choice of bioinformatic analysis pipeline alone can result in different estimations of the composition of the microbiome when using 16S rRNA gene amplicon sequencing data. The study observed differences in terms of both presence and abundance of organisms and provides a resource for ensuring reproducible pipeline development and application. The observed differences were especially prevalent when using custom databases and applying high stringency operational taxonomic unit (OTU) cut-off limits. In order to apply sequencing approaches with greater accuracy, the impact of different analytical steps needs to be clearly delineated and solutions devised to harmonise microbiome analysis results.

The application of high-throughput sequencing in profiling microbial communities is providing an unprecedented ability to investigate microbiomes. Such studies typically apply one of two methods: amplicon sequencing using PCR to target a conserved orthologous sequence (typically the 16S ribosomal RNA gene) or whole (meta)genome sequencing (WGS). Both methods have been used to catalog the microbial taxa present in a sample and quantify their respective abundances. However, a comparison of the inherent precision or bias of the different sequencing approaches has not been performed. We previously developed a metagenomic control material (MCM) to investigate error when performing different sequencing strategies. Amplicon sequencing using four different primer strategies and two 16S rRNA regions was examined (Roche 454 Junior) and compared to WGS (Illumina HiSeq). All sequencing methods generally performed comparably and in good agreement with organism specific digital PCR (dPCR); WGS notably demonstrated very high precision. Where discrepancies between relative abundances occurred they tended to differ by less than twofold. Our findings suggest that when alternative sequencing approaches are used for microbial molecular profiling they can perform with good reproducibility, but care should be taken when comparing small differences between distinct methods. This work provides a foundation for future work comparing relative differences between samples and the impact of extraction methods. We also highlight the value of control materials when conducting microbial profiling studies to benchmark methods and set appropriate thresholds.

Background: Antimicrobial resistance (AMR) poses a threat to public health. Clinical microbiology laboratories typically rely on culturing bacteria for antimicrobial susceptibility testing (AST). As the implementation costs and technical barriers fall, whole-genome sequencing (WGS) has emerged as a ‘one-stop’ test for epidemiological and predictive AST results. Few published comparisons exist for the myriad analytical pipelines used for predicting AMR. To address this, we performed an inter-laboratory study providing sets of participating researchers with identical short-read WGS data sequenced from clinical isolates, allowing us to assess the reproducibility of the bioinformatic prediction of AMR between participants and identify problem cases and factors that lead to discordant results. Methods: We produced ten WGS datasets of varying quality from cultured carbapenem-resistant organisms obtained from clinical samples sequenced on either an Illumina NextSeq or HiSeq instrument. Nine participating teams (‘participants’) were provided these sequence data without any other contextual information. Each participant used their own pipeline to determine the species, the presence of resistance-associated genes, and to predict susceptibility or resistance to amikacin, gentamicin, ciprofloxacin and cefotaxime. Results: Individual participants predicted different numbers of AMR-associated genes and different gene variants from the same clinical samples. The quality of the sequence data, choice of bioinformatic pipeline and interpretation of the results all contributed to discordance between participants. Although much of the inaccurate gene variant annotation did not affect genotypic resistance predictions, we observed low specificity when compared to phenotypic AST results but this improved in samples with higher read depths. Had the results been used to predict AST and guide treatment a different antibiotic would have been recommended for each isolate by at least one participant. Conclusions: We found that participants produced discordant predictions from identical WGS data. These challenges, at the final analytical stage of using WGS to predict AMR, suggest the need for refinements when using this technology in clinical settings. Comprehensive public resistance sequence databases and standardisation in the comparisons between genotype and resistance phenotypes will be fundamental before AST prediction using WGS can be successfully implemented in standard clinical microbiology laboratories.