Validation of an automated, end-to-end metagenomic sequencing assay for agnostic detection of respiratory viruses

BACKGROUND

Current molecular diagnostics are limited in the number and type of detectable pathogens. Metagenomic next generation sequencing (mNGS) is an emerging, and increasingly feasible, pathogen-agnostic diagnostic approach. Translational barriers prohibit the widespread adoption of this technology in clinical laboratories. We validate an end-to-end mNGS assay for detection of respiratory viruses. Our assay is optimized to reduce turnaround time, lower cost-per-sample, increase throughput, and deploy secure and actionable bioinformatic results.

METHODS

We validated our assay using residual nasopharyngeal swab specimens from Vancouver General Hospital (n = 359), RT-PCR-positive, or negative for Influenza, SARS-CoV-2, and RSV. We quantified sample stability, assay precision, the effect of background nucleic acid levels, and analytical limits of detection. Diagnostic performance metrics were estimated.

RESULTS

We report that our mNGS assay is highly precise, semi-quantitative, with analytical limits of detection ranging from 103-104 copies/mL. Our assay is highly specific (100%) and sensitive (61.9% Overall: 86.8%; RT-PCR Ct < 30). Multiplexing capabilities enable processing of up to 55-specimens simultaneously on an Oxford Nanopore GridION device, with results reported within 12-hours.

CONCLUSIONS

This study outlines the diagnostic performance and feasibility of mNGS for respiratory viral diagnostics, infection control, and public health surveillance. We addressed translational barriers to widespread mNGS adoption.

Ten common issues with reference sequence databases and how to mitigate them

Metagenomic sequencing has revolutionized our understanding of microbiology. While metagenomic tools and approaches have been extensively evaluated and benchmarked, far less attention has been given to the reference sequence database used in metagenomic classification. Issues with reference sequence databases are pervasive. Database contamination is the most recognized issue in the literature; however, it remains relatively unmitigated in most analyses. Other common issues with reference sequence databases include taxonomic errors, inappropriate inclusion and exclusion criteria, and sequence content errors. This review covers ten common issues with reference sequence databases and the potential downstream consequences of these issues. Mitigation measures are discussed for each issue, including bioinformatic tools and database curation strategies. Together, these strategies present a path towards more accurate, reproducible and translatable metagenomic sequencing.

WGS of a cluster of MDR Shigella sonnei utilizing Oxford Nanopore R10.4.1 long-read sequencing

OBJECTIVES

To utilize long-read nanopore sequencing (R10.4.1 flowcells) for WGS of a cluster of MDR Shigella sonnei, specifically characterizing genetic predictors of antimicrobial resistance (AMR).

METHODS

WGS was performed on S. sonnei isolates identified from stool and blood between September 2021 and October 2022. Bacterial DNA from clinical isolates was extracted on the MagNA Pure 24 and sequenced on the GridION utilizing R10.4.1 flowcells. Phenotypic antimicrobial susceptibility testing was interpreted based on CLSI breakpoints. Sequencing data were processed with BugSeq, and AMR was assessed with BugSplit and ResFinder.

RESULTS

Fifty-six isolates were sequenced, including 53 related to the cluster of cases. All cluster isolates were identified as S. sonnei by sequencing, with global genotype 3.6.1.1.2 (CipR.MSM5), MLST 152 and PopPUNK cluster 3. Core genome MLST (cgMLST, examining 2513 loci) and reference-based MLST (refMLST, examining 4091 loci) both confirmed the clonality of the isolates. Cluster isolates were resistant to ampicillin (blaTEM-1), trimethoprim/sulfamethoxazole (dfA1, dfrA17; sul1, sul2), azithromycin (ermB, mphA) and ciprofloxacin (gyrA S83L, gyrA D87G, parC S80I). No genomic predictors of resistance to carbapenems were identified.

CONCLUSIONS

WGS with R10.4.1 enabled rapid sequencing and identification of an MDR S. sonnei community cluster. Genetic predictors of AMR were concordant with phenotypic antimicrobial susceptibility testing.

No Isolate, No Problem: Using a Novel Insertion Sequence PCR to Link Rats to Human Shigellosis Cases in an Underserved Urban Community

During an investigation into a cluster of Shigella flexneri serotype 2a cases in an underserved community, we assessed the relatedness of human and rat S. flexneri isolates utilizing a novel PCR targeting insertion sites (IS-PCR) of mobile elements in the Shigella genome characteristic of the cluster strain. Whole-genome sequences of S. flexneri (n = 50) associated with the cluster were analyzed. De novo genome assemblies were analyzed by a Geneious V10.2.6 motif search, and two unique IS were identified in all human Shigella sequences of the local cluster. Hydrolysis probe PCR assays were designed to detect these sequences consisting of forward and reverse primers to amplify across each insertion site and a hydrolysis probe spanning the insertion site. IS-PCR was performed for three Shigella PCR-positive culture-negative rat intestine specimens from this community. Both insertion sites were detected in the de novo genome assemblies of all clinical S. flexneri isolates (n = 50). Two of the three PCR-positive culture-negative rat samples were positive for both unique ISs identified in the human S. flexneri isolates, suggesting that the rat Shigella species strains were closely related to the human strains in the cluster. The cycle threshold (Ct) values were >35, indicating that the bacterial load was very low in the rat samples. Two unique IS were identified in clinical isolates from a community S. flexneri cluster. Both IS targets were identified in PCR-positive (Shigella spp.), culture-negative rat tissue and clinical isolates from humans, indicating relatedness. IMPORTANCE This article describes a novel molecular method to show relatedness between bacterial infections, which may not be able to grow in the laboratory due to treatment with antibiotics or for bacteria requiring unique conditions to grow well. Uniquely, we applied this technique to Shigella isolates from human cases associated with a local cluster in an underserved community, as well as rat samples from the same community. We believe that this novel approach can serve as a complementary method to support outbreak/cluster investigation for Shigella spp.

Agnostic Sequencing for Detection of Viral Pathogens

The advent of next-generation sequencing (NGS) technologies has expanded our ability to detect and analyze microbial genomes and has yielded novel molecular approaches for infectious disease diagnostics. While several targeted multiplex PCR and NGS-based assays have been widely used in public health settings in recent years, these targeted approaches are limited in that they still rely on a priori knowledge of a pathogen's genome, and an untargeted or unknown pathogen will not be detected. Recent public health crises have emphasized the need to prepare for a wide and rapid deployment of an agnostic diagnostic assay at the start of an outbreak to ensure an effective response to emerging viral pathogens. Metagenomic techniques can nonspecifically sequence all detectable nucleic acids in a sample and therefore do not rely on prior knowledge of a pathogen's genome. While this technology has been reviewed for bacterial diagnostics and adopted in research settings for the detection and characterization of viruses, viral metagenomics has yet to be widely deployed as a diagnostic tool in clinical laboratories. In this review, we highlight recent improvements to the performance of metagenomic viral sequencing, the current applications of metagenomic sequencing in clinical laboratories, as well as the challenges that impede the widespread adoption of this technology.

Automated Library Construction and Analysis for High-Throughput Nanopore Sequencing of SARS-CoV-2

Background

To support the implementation of high-throughput pipelines suitable for SARS-CoV-2 sequencing and analysis in a clinical laboratory, we developed an automated sample preparation and analysis workflow.

Methods

We used the established ARTIC protocol with approximately 400 bp amplicons sequenced on Oxford Nanopore’s MinION. Sequences were analyzed using Nextclade, assigning both a clade and quality score to each sample.

Results

A total of 2179 samples on twenty-five 96-well plates were sequenced. Plates of purified RNA were processed within 12 h, sequencing required up to 24 h, and analysis of each pooled plate required 1 h. The use of samples with known threshold cycle (Ct) values enabled normalization, acted as a quality control check, and revealed a strong correlation between sample Ct values and successful analysis, with 85% of samples with Ct < 30 achieving a “good” Nextclade score. Less abundant samples responded to enrichment with the fraction of Ct > 30 samples achieving a “good” classification rising by 60% after addition of a post-ARTIC PCR normalization. Serial dilutions of 3 variant of concern samples, diluted from approximately Ct = 16 to approximately Ct = 50, demonstrated successful sequencing to Ct = 37. The sample set contained a median of 24 mutations per sample and a total of 1281 unique mutations with reduced sequence read coverage noted in some regions of some samples. A total of 10 separate strains were observed in the sample set, including 3 variants of concern prevalent in British Columbia in the spring of 2021.

Conclusions

We demonstrated a robust automated sequencing pipeline that takes advantage of input Ct values to improve reliability.

Nanopore metagenomic sequencing for detection and characterization of SARS-CoV-2 in clinical samples

OBJECTIVES

The COVID-19 pandemic has underscored the need for rapid novel diagnostic strategies. Metagenomic Next-Generation Sequencing (mNGS) may allow for the detection of pathogens that can be missed in targeted assays. The goal of this study was to assess the performance of nanopore-based Sequence-Independent Single Primer Amplification (SISPA) for the detection and characterization of SARS-CoV-2.

METHODS

We performed mNGS on clinical samples and designed a diagnostic classifier that corrects for barcode crosstalk between specimens. Phylogenetic analysis was performed on genome assemblies.

RESULTS

Our assay yielded 100% specificity overall and 95.2% sensitivity for specimens with a RT-PCR cycle threshold value less than 30. We assembled 10 complete, and one near-complete genomes from 20 specimens that were classified as positive by mNGS. Phylogenetic analysis revealed that 10/11 specimens from British Columbia had a closest relative to another British Columbian specimen. We found 100% concordance between phylogenetic lineage assignment and Variant of Concern (VOC) PCR results. Our assay was able to distinguish between the Alpha and Gamma variants, which was not possible with the current standard VOC PCR being used in British Columbia.

CONCLUSIONS

This study supports future work examining the broader feasibility of nanopore mNGS as a diagnostic strategy for the detection and characterization of viral pathogens.

Next-generation sequencing for cytomegalovirus antiviral resistance genotyping in a clinical virology laboratory

INTRODUCTION

The identification of CMV antiviral drug resistance (AVDR) is a critical diagnostic test for immunocompromised patients with CMV infection and a failure of virologic response on optimal antiviral treatment. We developed a next-generation sequencing (NGS) assay for CMV AVDR testing and compared the AVDR mutations identified by NGS to Sanger sequencing.

METHODS

Retrospective review of CMV AVDR testing requests for UL97 and UL54 at our laboratory from 2014 to 2019 was conducted. NGS was performed on the MinION and compared to Sanger sequencing performed at the national reference laboratory. Analysis of the sequences was completed with a novel cloud bioinformatics platform (BugSeq).

RESULTS

Twenty patient samples previously characterized were included for study on the MinION. NGS captured all of the CMV AVDR mutations identified by Sanger, and identified additional mutations in UL97 and/or UL54 in 8/13 (62%) of the samples. An analysis of the depth of coverage at which we no longer detected minority single nucleotide variants (SNVs) detected in the original data was conducted, estimating a recall of 95% at 1800 fold coverage.

CONCLUSION

NGS utilizing MinION technology for the detection of CMV AVDR mutations identified additional minority variants in UL97 and UL54 as compared with Sanger sequencing. Through the application of a bioinformatics pipeline available online, our NGS process eliminates barriers associated with the use of the MinION and NGS in clinical laboratories.

Reanalysis of Alzheimer's brain sequencing data reveals absence of purported HHV6A and HHV7

Readhead et al. recently reported in Neuron the detection and association of human herpesviruses 6A (HHV6A) and 7 (HHV7) with Alzheimer's disease by shotgun sequencing. I was skeptical of the specificity of their modified Viromescan bioinformatics method and subsequent analysis for numerous reasons. Using their supplementary data, the prevalence of variola virus, the etiological agent of the eradicated disease smallpox, can be calculated at 97.5% of their Mount Sinai Brain Bank dataset. Reanalysis of Readhead et al.'s data using highly sensitive and specific alternative methods finds no HHV7 reads in their samples; HHV6A reads were found in only 2 out of their top 15 samples sorted by reported HHV6A abundance. Finally, recreation of Readhead et al.'s modified Viromescan method identifies reasons for its low specificity.

BugSeq: a highly accurate cloud platform for long-read metagenomic analyses

Background

As the use of nanopore sequencing for metagenomic analysis increases, tools capable of performing long-read taxonomic classification (ie. determining the composition of a sample) in a fast and accurate manner are needed. Existing tools were either designed for short-read data (eg. Centrifuge), take days to analyse modern sequencer outputs (eg. MetaMaps) or suffer from suboptimal accuracy (eg. CDKAM). Additionally, all tools require command line expertise and do not scale in the cloud.

Results

We present BugSeq, a novel, highly accurate metagenomic classifier for nanopore reads. We evaluate BugSeq on simulated data, mock microbial communities and real clinical samples. On the ZymoBIOMICS Even and Log communities, BugSeq (F1 = 0.95 at species level) offers better read classification than MetaMaps (F1 = 0.89–0.94) in a fraction of the time. BugSeq significantly improves on the accuracy of Centrifuge (F1 = 0.79–0.93) and CDKAM (F1 = 0.91–0.94) while offering competitive run times. When applied to 41 samples from patients with lower respiratory tract infections, BugSeq produces greater concordance with microbiological culture and qPCR compared with “What’s In My Pot” analysis.

Conclusion

BugSeq is deployed to the cloud for easy and scalable long-read metagenomic analyses. BugSeq is freely available for non-commercial use at https://bugseq.com/free.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12859-021-04089-5.

Phenotypic and Genotypic Correlates of Penicillin Susceptibility in Nontoxigenic Corynebacterium diphtheriae, British Columbia, Canada, 2015-2018

In 2015, the Clinical and Laboratory Standards Institute (CLSI) updated its breakpoints for penicillin susceptibility in Corynebacterium species from <1 mg/L to <0.12 mg/L. We assessed the effect of this change on C. diphtheriae susceptibility reported at an inner city, tertiary care center in Vancouver, British Columbia, Canada, during 2015–2018 and performed whole-genome sequencing to investigate phenotypic and genotypic resistance to penicillin. We identified 44/45 isolates that were intermediately susceptible to penicillin by the 2015 breakpoint, despite meeting previous CLSI criteria for susceptibility. Sequencing did not reveal β-lactam resistance genes. Multilocus sequence typing revealed a notable predominance of sequence type 76. Overall, we saw no evidence of penicillin nonsusceptibility at the phenotypic or genotypic level in C. diphtheriae isolates from our institution. The 2015 CLSI breakpoint change could cause misclassification of penicillin susceptibility in C. diphtheriae isolates, potentially leading to suboptimal antimicrobial treatment selection.

Adjunctive bradyzoite-directed therapy for reducing complications of congenital toxoplasmosis

Congenital toxoplasmosis is caused by in utero infection of the fetus with the intracellular parasite Toxoplasma gondii. Upon infection, the parasite forms life-long cysts in fetal brain and eyes which are resistant to the currently accepted therapy of pyrimethamine and sulfadiazine. These cysts commonly reactivate later in life causing chorioretinitis and visual impairment, and rarely cause neurological complications. I hypothesize that adjunctive, bradyzoite-directed therapies have the potential to alleviate a significant burden of disease by reducing cyst burden in neonatal brain and eyes. Atovaquone is perhaps the most promising drug for further evaluation given its low side-effect profile, established safety, and efficacy in animal models reducing cyst burden. Very limited observational data in humans suggests atovaquone may prevent Toxoplasma-associated chorioretinitis recurrence. Clinical trials are needed to evaluate it and other potential drugs as adjunctive treatment in congenital toxoplasmosis.