The Society for Family Health (SFH) Nigeria is a non governmental organisation (NGO), incorporated in 1985, focused on providing malaria prevention and treatment (including intermittent preventive treatment), HIV prevention (including Prevention of ), maternal and child health, sexual and reproductive health, family planning, cervical cancer screening and prevention and safe water systems. SFH uses social marketing, behaviour change communication (communication through radio dramas, mass media messages, etc.) and research working in partnership with the Government of Nigeria and community-based organisations. Contents.History SFH is a public health institution in Nigeria founded in 1983 by Honourable Justice Ifeyinwa Nzeako, Prof. Olikoye Ransome-Kuti (late), Pharm (Mallam) Dahiru Wali and Mr. SFH began as a (PSI) affiliate with one HIV grant and reproductive health products for distribution in Nigeria.In 1985 SFH was incorporated as a Nigerian non-governmental organisation and in 1994 released the award-winning 'Who Get This Rain Coat' Gold Circle Condom/Family Planning campaign on national television.

Simply put, journalism needs the likes of Mr. Olokor to advance the frontiers of democracy and accountability, so that society will be rescued from the shackles of injustice and deprivation facing Nigeria. That aside, Olokor was born on November 22, 1968, in Abavo, Ika South Local Government Area of Delta. Of the terminally ill toward impending death (Olokor, 1998), and it is the culture of a people that greatly determines their attitudes toward death and dying. Typically, the terminally ill hold a personal hope of overcoming their diseased state, and personal efforts are made to maintain life (e.g., strictly following all instructions given.

In the late 1980s it went into partnerships with pharmaceutical companies to distribute Gold Circle condoms in Lagos, Oyo and Ogun States, in South Western Nigeria. It soon scaled up its operations, expanding nationwide, and commencing the marketing of oral contraceptives, in partnership with. A partnership with the UK's (DFID) followed, which also focuses on the marketing of condoms oral and injectable contraceptives, and a water-based lubricant.By 1997 SFH was distributing seventeen million condoms annually, and by 2009, 200 million condoms all over Nigeria at a subsidised price. In 2003 SFH launched its malaria programme, in partnership with USAID.

This focused on both the treatment and prevention of malaria, one of the biggest causes of infant and child mortality in Nigeria. The malaria programme has grown significantly, with funding and support from. The focus is now on the use of (ACT) and conducting Rapid Diagnostic Test for malaria before treatment as well as encouraging proper use of Long Lasting Insecticidal Nets.In 2005 SFH became the first Nigerian organisation to receive direct funding from the United States Agency for International Development (USAID) to implement programmes in reproductive health., by; at; published 1 December 2013; retrieved 16 April 2014., by Remi Oyo; at; published 12 November 2000; retrieved 16 April 2014. 2014-05-31 at the, by Friday Olokor; at; published 15 January 2014; retrieved 16 April 2014., by Seyi Ogunbameru; at; published 23 August 2013; retrieved 16 April 2014.

2012-07-19 at the, by Wale Adepoju; at; published 23 March 2012; retrieved 16 April 2014., by 'Sola Fagorusi; at; published 17 April 2013; retrieved 16 April 2014., by USAID; at; published 2 June 2006; retrieved 16 April 2014. Retrieved 2017-10-24. Retrieved 2017-10-24.

Influenza is a major global public health threat as a result of its highly pathogenic variants, large zoonotic reservoir, and pandemic potential. Metagenomic viral sequencing offers the potential for a diagnostic test for influenza virus which also provides insights on transmission, evolution, and drug resistance and simultaneously detects other viruses. We therefore set out to apply the Oxford Nanopore Technologies sequencing method to metagenomic sequencing of respiratory samples. We generated influenza virus reads down to a limit of detection of 10 2 to 10 3 genome copies/ml in pooled samples, observing a strong relationship between the viral titer and the proportion of influenza virus reads ( P = 4.7 × 10 −5). Applying our methods to clinical throat swabs, we generated influenza virus reads for 27/27 samples with mid-to-high viral titers (cycle threshold C T values, 99% complete sequences for all eight gene segments.

We also detected a human coronavirus coinfection in one clinical sample. While further optimization is required to improve sensitivity, this approach shows promise for the Nanopore platform to be used in the diagnosis and genetic analysis of influenza virus and other respiratory viruses. INTRODUCTIONInfluenza A virus is an RNA orthomyxovirus of approximately 13 kb in length, with an eight-segment genome. It is typically classified on the basis of hemagglutinin (HA) and neuraminidase (NA), of which there are 16 and 9 main variants, respectively. Genetic reassortment underpins the potential for transmission between different host species and for the evolution of highly pathogenic variants (,), recognized in the WHO list of “ten threats to global health”. Seasonal influenza causes an estimated 650,000 deaths globally each year, and the H3N2 variant alone kills 35,000 people each year in the United States (, ). Certain groups are particularly at risk, including older adults, infants, young children, pregnant women, those with underlying lung disease, and the immunocompromised.

The burden of disease disproportionately affects low-/middle-income settings. Influenza virus diagnostics and surveillance are fundamental to identify the emergence of novel strains, to improve the prediction of potential epidemics and pandemics (, ), and to inform vaccine strategy. Diagnostic data facilitate real-time surveillance, can underpin infection control interventions (, ), and can inform the prescription of neuraminidase inhibitors (NAI).Currently, most clinical diagnostic tests for influenza virus depend on detecting viral antigen or on PCR amplification of viral nucleic acid derived from respiratory samples. These two approaches offer trade-offs in benefits, as follows: antigen tests (including point-of-care tests POCT) are typically rapid but have low sensitivity (,), while PCR is more time-consuming but more sensitive. Irrespective of the test used, most clinical diagnostic facilities report a nonquantitative (binary) diagnostic result, and the data routinely generated for influenza diagnosis have limited capacity to inform insights into epidemiological linkage, vaccine efficacy, or antiviral susceptibility.

On these grounds, there is an aspiration to generate new diagnostic tests that combine speed (incorporating the potential for POCT , ), sensitivity, detection of coinfection (, ), and generation of quantitative or semiquantitative data that can be used to identify drug resistance and reconstruct phylogeny to inform surveillance, public health strategy, and vaccine design.The application of Oxford Nanopore Technologies (ONT) sequencing to generate full-length influenza virus sequences from clinical respiratory samples can address these challenges. ONT offers a “third-generation,” portable, real-time approach to generating long-read single-molecule sequence data, with demonstrated success across a range of viruses (,).

To date, Nanopore sequencing of influenza virus has been reported using high-titer virus from an in vitro culture system, producing full-length genome sequences through direct RNA sequencing , or using targeted enrichment by either hybridization of cDNA or influenza virus-specific PCR amplification.We therefore aimed to optimize a metagenomic protocol for detecting influenza viruses directly from clinical samples using Nanopore sequencing. We determine its sensitivity compared to that of existing diagnostic methods and its accuracy compared to short-read (Illumina) sequencing, using clinical samples from hospital patients during an influenza season and samples from a controlled laboratory infection in ferrets.

Further optimization is required before the Nanopore method can be rolled out as a diagnostic test, but we highlight the potential impact of this technology in advancing molecular diagnostics for respiratory pathogens. Schematic to show processing protocol through clinical and research pipelines for influenza diagnosis. (A) Clinical sample collection (orange), clinical diagnostic testing (yellow), sample processing and sequencing using Oxford Nanopore Technologies (blue), and processing of sequence data (purple). (B) Outline of pooled influenza virus-positive samples into an influenza virus-negative background to generate various titers of influenza virus (from 0 to 10 6 genome copies/ml), undertaken in triplicate, and spiked with a standard titer of Hazara virus control at 10 4 genome copies/ml. FluA, influenza A virus.For methodological assessment, we focused on four categories of samples, as follows: positive pool, negative pools, individual positive samples, and individual negative samples. For the positive pool, we pooled 19 throat swab samples that had tested positive for influenza A virus in the clinical diagnostic laboratory to provide a large enough sample to assess reproducibility. For the negative pools, we generated three pools of throat swab samples that had tested negative for influenza virus (consisting of 24, 38, and 38 individual samples).

For the individual positive samples, we included 40 individual samples (35 throat swabs and 5 nasal swabs) that had tested positive for influenza A or B virus, selected to represent the widest range of GeneXpert assay C T values (13.5 to 39.3; valid test result range, 12 to 40). Gunz the duel k style 5. For the individual negative samples, we selected 10 individual throat swab samples that were influenza virus negative.

Optimization of methods.Prior to establishing the protocol detailed in full below, we assessed the impact of two possible optimization steps, centrifugation versus filtration and reduced time for cDNA synthesis. For centrifugation versus filtration, we investigated two approaches to deplete human/bacterial nucleic acid from our samples, i.e., filtration of the raw sample via a 0.4-μm filter (Sartorius) before further processing versus using a hard spin (16,000 × g for 2 min). CDNA libraries for this comparison were produced as described previously. For the reduced time for cDNA synthesis, to assess the possibility of time saving in the cDNA synthesis steps, we compared performance of the previously described protocol to that of a modified version with two alterations, first using SuperScript IV (Thermo Fisher) in place of SuperScript III (Thermo Fisher) for reverse transcription, with the incubation time reduced from 60 min to 10 min at 42°C, and second, reducing the cDNA amplification PCR extension cycling time from 5 min to 2 min. Positive control.Prior to nucleic acid extraction, each sample was spiked with Hazara virus virions to a final concentration of 10 4 genome copies per ml as a positive internal control.

This is an enveloped negative-stranded RNA virus (genus Orthonairovirus, order Bunyavirales) with a trisegmented genome of 11,980, 4,575, and 1,677 nucleotides in length (GenBank accession numbers to ). It is nonpathogenic in humans and would therefore not be anticipated to arise in any of our clinical samples.

Cultured virions from an SW13 cell line were provided by the National Collection of Pathogenic Viruses (NCPV; catalog no. Nucleic acid extraction.Samples were centrifuged at 16,000 × g for 2 min. The supernatant was eluted without disturbing the pelleted material and was used in nucleic acid extraction. Total nucleic acid was extracted from 100 μl of supernatant using the QIAamp viral RNA kit (Qiagen) eluting in 50 μl of H 2O, followed by a DNase treatment with Turbo DNase (Thermo Fisher Scientific) at 37°C for 30 min.

RNA was purified and concentrated to 6 μl using the RNA Clean & Concentrator-5 kit (Zymo Research), following the manufacturer’s instructions. Randomly amplified cDNA was prepared for each sample using a sequence-independent single-primer amplification (SISPA) approach, adapted from our previously described workflow , based on the round A/B methodology. For reverse transcription, 4 μl of RNA and 1 μl of primer A (5′-GTTTCCCACTGGAGGATA-N9-3′, 40 pmol/μl) were mixed and incubated for 5 min at 65°C and then cooled to room temperature. First-strand synthesis was performed by the addition of 2 μl SuperScript IV first-strand buffer, 1 μl of 12.5 mM dinucleoside triphosphates (dNTPs), 0.5 μl of 0.1 M dithiothreitol (DTT), 1 μl H 2O, and 0.5 μl SuperScript IV (Thermo Fisher) before incubation for 10 min at 42°C.

Second-strand synthesis was performed by the addition of 1 μl Sequenase buffer, 3.85 μl H 2O, and 0.15 μl Sequenase (Affymetrix) prior to incubation for 8 min at 37°C, followed by the addition of 0.45 μl Sequenase dilution buffer and 0.15 μl Sequenase and a further incubation at 37°C for 8 min. Amplification of cDNA was performed in triplicate using 5 μl of the reaction mixture as input to a 50-μl AccuTaq LA (Sigma) reaction mixture, according to the manufacturer’s instructions, using 1 μl primer B (5′-GTTTCCCACTGGAGGATA-3′) , with PCR cycling conditions of 98°C for 30 s, 30 cycles of 94°C for 15 s, 50°C for 20 s, and 68°C for 2 min, followed by 68°C for 10 min. Amplified cDNA was pooled from the triplicate reaction mixtures, purified using a 1:1 ratio of AMPure XP beads (Beckman Coulter, Brea, CA), and quantified using a Qubit high-sensitivity double-stranded DNA (dsDNA) kit (Thermo Fisher), both according to the manufacturers’ instructions.

Nanopore library preparation and sequencing.Multiplex sequencing libraries were prepared using 250 ng of cDNA from up to 12 samples as input to the SQK-LSK108 or SQK-LSK109 kit and barcoded individually using the EXP-NBD103 Native barcodes (Oxford Nanopore Technologies) and a modified One-pot protocol. Libraries were sequenced on FLO-MIN106 flow cells on the MinION Mk1b or GridION device (Oxford Nanopore Technologies), with sequencing proceeding for 48 h. Samples were batched according to the GeneXpert C T value (see File S1 in the supplemental material). Bioinformatic analysis.Nanopore reads were base called using Guppy (Oxford Nanopore Technologies, Oxford, UK). Output base called fastq files were demultiplexed using Porechop v0.2.3.

The reads were first taxonomically classified against the RefSeq database using Centrifuge v1.0.3. The reads were then mapped against the reference sequence selected from the Centrifuge report using Minimap2 v2.9 (, ). A draft consensus sequence was generated by using a majority voting approach to determine the nucleotide at each position. The resulting draft consensus sequences were subjected to a BLAST search against an influenza virus sequence database that included 2,000 H1N1 and H3N2 seasonal influenza virus sequences between 2018 and 2019 and were downloaded from the Influenza Research Database. The reads were again mapped against the reference sequence using Minimap2 v2.9, and the number of mapped reads was calculated using SAMtools v1.5 and Pysam. The subtype of the influenza A virus derived from each clinical sample was determined by the subtypes of the HA and NA reference sequences.

A consensus sequence was built using Nanopolish v0.11.0 (, ) and the margincons.py script. For the Illumina data, reads were quality trimmed to a minimum score of Q30 across the read with Trimmomatic. BWA-MEM v0.7.15 was used to align the reads to reference genomes using MEM defaults. SAMtools v1.4 was used to compute the percent reads mapped and coverage depth. Mapping consensuses for Illumina sequencing were generated using QuasiBam. Maximum likelihood phylogeny was generated for the HA gene segment using RAxML v8.2.10 , in which a general time-reversible model of nucleotide substitution and a gamma-distributed rate variation among sites were applied. Sequence alignments were performed by using MUSCLE v3.8.

Method optimization to increase the proportion of viral reads derived from throat swabs.Our method protocol is shown in. We first sequenced five influenza A virus-positive and five influenza virus-negative throat swabs, each spiked with Hazara virus control at 10 4 genome copies/ml. Using a sequence-independent single-primer amplification (SISPA) approach , followed by Nanopore sequencing, we produced metagenomic data dominated by reads that were bacterial in origin, with extremely few viral reads detected.

Passing the sample through a 0.4-μm filter prior to nucleic acid extraction increased the detection of viral reads by several orders of magnitude (Fig. Filtration is relatively expensive, so we also assessed the alternative approach of adding a rapid-centrifugation step to pellet bacterial and human cells, followed by nucleic acid extraction from the supernatant.

We used a pooled set of influenza A virus-positive samples (concentration, 10 6 genome copies/ml) to provide a large enough sample to assess reproducibility, with the Hazara virus control spiked in at 10 4 genome copies/ml. Enrichment for influenza virus and Hazara virus was similar for filtration versus centrifugation, based on read mapping to the viral genome (Fig. As centrifugation is simpler and less expensive, we selected this approach for all further testing. Method optimization to reduce time for cDNA synthesis.Synthesis of tagged randomly primed cDNA and its subsequent amplification via SISPA required lengthy reverse transcription and PCR steps (1 h and 3 h 45 min), respectively. Optimizing these stages upgraded the reverse transcriptase from SuperScript III to SuperScript IV (Thermo Fisher), reduced the incubation time to 10 min (processing time reduction, 50 min), and reduced the PCR extension time within each cycle from 5 min to 2 min (1 h 30 min processing time reduction). Comparing this final method with our original protocol, using triplicate extractions from the pooled set of influenza A virus-positive samples demonstrated no significant loss in performance in the more rapid protocol (Fig. S3), and we adopted this approach as our routine protocol, giving a wet-lab processing time of ∼8 h.

Consistent retrieval of Hazara virus by Nanopore sequencing.Starting with an influenza A virus-positive sample pool (10 6 genome copies/ml), we made three volumetric dilution series using three independent influenza virus-negative pools. The total quantity of cDNA after preparation for sequencing was consistently higher in all samples using negative pool 3 as the diluent , indicating the presence of a higher concentration of nonviral RNA within pool 3. This is likely due to host cell lysis or higher bacterial presence and demonstrates the variable nature of throat swab samples. Characteristics of three pools of influenza virus-negative throat swabs and Nanopore sequence results following spiking with influenza A virus.

(A) Total concentration of cDNA produced per pooled sample following amplification by the SISPA reaction, grouped by dilution series. The 10 6 genome copies/ml sample in each pool is the original, undiluted material, represented by the black bars. Samples diluted to influenza virus titers of 10 4, 10 3, and 10 2 contain more cDNA due to higher background material (bacterial/human) present in the diluent. Dilution series 1 and 2 contain comparable amounts of background material; dilution series 3 contains substantially more background. (B) Viral reads generated by Nanopore sequencing of samples with different titers of influenza A virus and a consistent titer of Hazara virus (10 4 genome copies/ml). Graphs show reads per million of total reads mapping to influenza A or Hazara virus genomes, across the three individual dilution series. Note the logarithmic scale on the y axis.We consistently retrieved Hazara virus reads from all three dilution series by Nanopore sequencing, independently of influenza virus titer in the sample.

Sequencing from dilution series 1 and 2 gave a consistent proportion of total reads mapping to the Hazara virus genome, across dilutions and between the first two pools, with mean ± standard deviation values per pool of 1.4 × 10 3 ± 660 reads per million (RPM) of total reads and 1.2 × 10 3 ± 350 RPM, respectively. The pool 3 dilution series generated 260 ± 340 RPM Hazara virus reads across samples and showed a decreasing trend associated with increased dilution factor as increasingly more nonviral RNA was introduced from this high-background pool. Limit of influenza virus detection by Nanopore sequencing from pooled samples.Nanopore sequencing of the triplicate SISPA preparations of the influenza A virus-positive pool produced mean ± standard deviation of 5.3 × 10 4 ± 3.6 × 10 4 RPM mapping to the influenza A virus genome. Across the dilution series, the proportion of influenza virus reads was strongly associated with influenza virus titer ( P value = 4.7 × 10 −5) but was also influenced by which negative pool was used for dilution, consistent with the pattern observed for the Hazara virus control.

Sequencing the negative controls (pools with no influenza virus spike) generated no reads mapping to influenza virus. At influenza virus titers of.

Retrieval and reconstruction of complete influenza virus genomes from pooled/spiked samples.For the Hazara virus control (10 4 genome copies/ml spike), genome coverage was 81.4 to 99.4% (at 1× depth) for pools 1 and 2. Coverage in the high-background pool 3 was more varied (21.5 to 96.5%; ).

Influenza A virus genome coverage at 10 6 copies/ml was ≥99.3% for each segment in all samples. At 10 4 genome copies/ml of influenza virus, a mean 1× coverage per segment was 90.3% for pools 1 and 2 but was substantially reduced in the high-background pool 3 to 5.7%.

At influenza virus titers of. Coverage of influenza virus and Hazara virus genome segments achieved by Nanopore sequencing from pooled samples (A) Data from three dilution series of pooled influenza virus-positive samples, diluted with three separate negative-sample pools to generate different titers of influenza virus. Each individual dilution was spiked with Hazara virus at 10 4 genome copies/ml. The proportion of genome covered at 1× depth is shown for each of the eight influenza virus genome segments (encoding PB2 polymerase subunit 2, PB1 polymerase subunit 1, PA polymerase acidic protein, HA hemagglutinin, NP nucleocapsid protein, NA neuraminidase, M matrix protein, and NS nonstructural protein) across the three dilution series. For simplicity, the coverage of the Hazara virus genome is plotted as the total of all three genome segments. (B) Representative coverage plots of influenza A virus genome segments from the dilution series 1 sample at 10 4 influenza virus copies per ml. Influenza A virus titer (genome copies/ml)Dilution pool no.Total no.

Of influenza A virus reads (reads per million)Influenza A virus subtyping bNo. Of Hazara virus reads (reads per million)10 3,103 (6.9 × 10 4)H3N2527 (1.1 × 10 3)2572,1066,957 (1.2 × 10 4)H3N2102 (178)3526,85241,196 (7.8 × 10 4)H3N2534 (1.0 × 10 3)10 80 (791)H3N2738 (2.1 × 10 3)2433,033299 (690)H3N2691 (1.6 × 10 3)343,5122 (46)Not possible9 (21,9297 (30)H3N2298 (1.3 × 10 3)2461,28124 (52)H3N2638 (1.4 × 10 3)3397,6722 (5)Not possible38 (5,1831 (3)Not possible453 (1.2 × 10 3)2671,1330 (0)Not possible598 (891)337,8970 (0)Not possible2 (53)Negative1903,4300 (0)NA1,731 (1.9 × 10 3)2900,4710 (0)NA692 (768)3818,5490 (0)NA54 (66). Influenza virus detection from individual clinical samples.Having demonstrated our ability to retrieve influenza virus sequences from pooled influenza virus-positive material diluted with negative samples, we next applied our methods to individual anonymized clinical samples, with 40 samples testing influenza virus positive and 10 samples testing influenza virus negative in the clinical diagnostic laboratory.

Data yield varied between flow cells (range, 2.5 × 10 6 to 13.2 × 10 6 reads from up to 12 multiplexed samples). Within flow cells, barcode performance was inconsistent when using a stringent, dual-barcode, demultiplexing method. From each clinical sample, the range of total reads generated was 1.0 × 10 5 to 2.4 × 10 6 (median, 3.8 × 10 5 reads) (Table S1).Reads mapping to either the influenza A or B virus genome were present in all 27 samples with a C T of 30, 6/13 samples generated influenza virus reads (range, 6 to 92,057 reads) (difference between sensitivity at a C T threshold of 30, P 90% of an influenza virus genome sequence was generated was 27.5 (Fig. Detection of Hazara virus internal control.Detection of the control virus (Hazara virus at 10 4 genome copies/ml) was highly varied, demonstrating that levels of background nontarget RNA are a major source of intersample variation. The number of Hazara virus reads per sample ranged from 0 to 13.5 × 10 3 (0 to 3.5 × 10 4 RPM), with a median of 70 reads (160 RPM) and mean of 706 reads (1.7 × 10 3 RPM) (Table S1). Four (8%) of 50 samples generated no detectable Hazara virus reads, two with high numbers of influenza virus reads (for sample 1, C T of 13.5 and 1.5 × 10 5 influenza B virus reads, and for sample 6, C T of 18.4 and 1.5 × 10 4 influenza A virus reads) acting to dilute the control signal. The other two samples contained no detectable influenza virus reads (for sample 34, C T of 35.9, and for sample 46, influenza virus negative).

The lack of control detection therefore indicates a loss of assay sensitivity due to high levels of background nucleic acid present in some samples. Comparison of Nanopore and Illumina sequencing.We selected a subset of 15 samples from across the viral titer range and resequenced on an Illumina MiSeq platform. The proportions of reads generated that mapped to the influenza virus genome were similar between the two sequencing technologies (Fig.

From 4 of the samples, nearly complete genomes were obtained. A comparison of consensus sequences derived from Nanopore and Illumina sequencing showed 100% concordance, except one sample that showed 7 nucleotide differences (identity, 99.94%) (Table S2). Detection of other RNA viruses in clinical samples.Within the 50 clinical samples sequenced, we found limited evidence for the presence of other RNA viruses. Sample 6 produced 109 reads mapping to human coronavirus in addition to 1.5 × 10 4 influenza A virus reads, suggesting coinfection.

We also derived 4.0 × 10 4 reads from human metapneumovirus from an influenza virus-negative sample, providing a nearly complete genome (99.8% coverage) from one sample (Fig. S1, sample I), further detailed previously. Animal time course study.Finally, we used samples collected from a previous animal experiment to test the reproducibility of our methods across a time course model of influenza A virus infection (three ferrets swabbed preinfection day −3 and then sampled at days 1, 2, 3, and 5 following laboratory infection with influenza A virus). The proportion of viral reads present at each time point was highly congruent with viral titer (titer is shown in and sequencing reads in ). We generated consensus genome sequences from Nanopore data at days 2, 3, and 5 postinfection; these were 100% concordant with Illumina-derived consensus sequences from the same cDNA (Table S2). DISCUSSIONTo our knowledge, this is the first report of successfully applying metagenomic Nanopore sequencing directly to respiratory samples to detect influenza virus and generate influenza virus sequences.

The approach demonstrates excellent specificity. Sensitivity varies by viral titer but is comparable to that of existing laboratory diagnostic tests for C T values of 90% complete genomes for 17/27 samples with a C T value of ≤30 (Fig. S4), demonstrating the ability of metagenomics to produce sufficient data for influenza virus diagnostics and genome characterization, while also detecting and sequencing other common RNA viruses.Despite time reductions in wet-laboratory processing, this method requires further modification to simplify and accelerate the protocol if it is to become viable as a near-to-patient test. High error rates are a recognized concern in Nanopore sequence data, and cross-barcode contamination can create challenges when low- and high-titer samples are batched. To avoid these problems, we batched samples according to C T value and applied stringent barcode demultiplexing criteria; however, this reduces the total data available for analysis, typically by ∼50% but with variation between sequencing runs. For future primary diagnostic use, it would be preferable to sequence samples individually using a lower-throughput flow cell, e.g., ONT Flongle (each paired with a negative-extraction-control sample, for which a prior spike with Hazara virus, using the same methods described here, would remain appropriate). Careful optimization of laboratory and bioinformatic methods is required to resolve individual sequence polymorphisms, particularly for drug resistance alleles.Infectious Diseases Society of America (IDSA) guidelines recommend nasal/nasopharyngeal specimens for influenza diagnosis, but throat swabs are easier to collect in clinical practice and therefore account for the majority of diagnostic samples processed by our clinical microbiology laboratory.

Further work is needed to investigate the sensitivity and specificity of our protocol for a wider array of respiratory sample types (also including bronchoalveolar lavage fluid, sputum, and saliva), which may contain different degrees of contaminating bacterial and/or human reads. Loss of assay sensitivity due to the presence of high-level background DNA from either the host or bacterial origin is a fundamental issue for metagenomic approaches, even in cell-free sample types such as cerebrospinal fluid. This challenge is exacerbated in throat swabs, as seen in our data. Our use of Hazara virus as an internal positive control allows us to identify those samples in which sensitivity has dropped to. Limitations of the method.The current limitations of metagenomic methods are their sensitivity in the context of low-pathogen-titer samples.

PCR-based methods measure the absolute count of viral genome copies present within a sample. Metagenomic sequencing measures the proportion of total RNA that is viral. Metagenomic sequencing is therefore affected by the level of nontarget RNA within a given sample, whereas PCR is not. As demonstrated here ( and ), detection of as little as 10 2 genome copies per ml is possible from throat swab samples (a level comparable with PCR-based methods), but variation in the level of background nucleic acids between individual samples makes detection at this level inconsistent. Further development of methods to deplete host and bacterial RNA within the samples is required to improve the performance of the assay at C T values of 30. Enrichment of pathogen sequences within libraries through either target capture or amplification is also an effective method to reduce the limit of target detection (, ) but requires the same a priori knowledge of both which pathogens are to be targeted and the full range of circulating viral diversity as other targeted methods discussed above, albeit with increased tolerance for diversity over PCR-based methods.

A further limitation compared to alternative sequencing technologies is the lack of confidence in determining the presence of minority variants due to the limited per-read accuracy, although we expect this to be addressed in future iterations of the ONT sequencing.In summary, while substantial further work is needed, our methods show promise for generating influenza virus sequences directly from respiratory samples. The “pathogen-agnostic” metagenomic sequencing approach offers an opportunity for simultaneous testing for a wide range of potential pathogens, providing a faster route to optimum treatment and contributing to antimicrobial stewardship. Longer term, this approach has promise as a routine laboratory test, providing data to inform treatment, vaccine design and deployment, infection control policies, and surveillance. The study was funded by the NIHR Oxford Biomedical Research Centre. Computation used the Oxford Biomedical Research Computing (BMRC) facility, a joint development between the Wellcome Centre for Human Genetics and the Big Data Institute supported by Health Data Research UK and the NIHR Oxford Biomedical Research Centre.The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health Research, the Department of Health, or Public Health England.P.C.M. Is funded by the Wellcome Trust (grant 110110). D.W.C., T.E.A.P., and A.S.W.

Are NIHR senior investigators. Iuliano AD, Roguski KM, Chang HH, Muscatello DJ, Palekar R, Tempia S, Cohen C, Gran JM, Schanzer D, Cowling BJ, Wu P, Kyncl J, Ang LW, Park M, Redlberger-Fritz M, Yu H, Espenhain L, Krishnan A, Emukule G, van Asten L, Pereira da Silva S, Aungkulanon S, Buchholz U, Widdowson M-A, Bresee JS, Global Seasonal Influenza-associated Mortality Collaborator Network. 2018.Estimates of global seasonal influenza-associated respiratory mortality: a modelling study.

Lancet 391:1285–1300. Quick J, Grubaugh ND, Pullan ST, Claro IM, Smith AD, Gangavarapu K, Oliveira G, Robles-Sikisaka R, Rogers TF, Beutler NA, Burton DR, Lewis-Ximenez LL, de Jesus JG, Giovanetti M, Hill SC, Black A, Bedford T, Carroll MW, Nunes M, Alcantara LC Jr, Sabino EC, Baylis SA, Faria NR, Loose M, Simpson JT, Pybus OG, Andersen KG, Loman NJ.2017.Multiplex PCR method for MinION and Illumina sequencing of Zika and other virus genomes directly from clinical samples.

Nat Protoc 12:1261–1276.