Next Article in Journal
The Immunomodulatory Role of Regulatory T Cells in Preterm Birth and Associated Pregnancy Outcomes
Next Article in Special Issue
Molecular Composition and Kinetics of B Cells During Ibrutinib Treatment in Patients with Chronic Lymphocytic Leukemia
Previous Article in Journal
Linking Antibodies Against Apolipoprotein A-1 to Metabolic Dysfunction-Associated Steatohepatitis in Mice
Previous Article in Special Issue
The Next Frontier in Tuberculosis Investigation: Automated Whole Genome Sequencing for Mycobacterium tuberculosis Analysis
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Single Laboratory Evaluation of the Q20+ Nanopore Sequencing Kit for Bacterial Outbreak Investigations

Genomics Development and Applications Branch, Division of Food Safety Genomics, Office of Applied Microbiology and Technology, Office of Laboratory Operations and Applied Science, Human Foods Program, Food & Drug Administration, College Park, MD 20740, USA
*
Author to whom correspondence should be addressed.
Int. J. Mol. Sci. 2024, 25(22), 11877; https://doi.org/10.3390/ijms252211877
Submission received: 27 September 2024 / Revised: 25 October 2024 / Accepted: 31 October 2024 / Published: 5 November 2024

Abstract

:
Leafy greens are a significant source of produce-related Shiga toxin-producing Escherichia coli (STEC) outbreaks in the United States, with agricultural water often implicated as a potential source. Current FDA outbreak detection protocols are time-consuming and rely on sequencing methods performed in costly equipment. This study evaluated the potential of Oxford Nanopore Technologies (ONT) with Q20+ chemistry as a cost-effective, rapid, and accurate method for identifying and clustering foodborne pathogens. The study focuses on assessing whether ONT Q20+ technology could facilitate near real-time pathogen identification, including SNP differences, serotypes, and antimicrobial resistance genes. This pilot study evaluated different combinations of two DNA extraction methods (Maxwell RSC Cultured Cell DNA kit and Monarch high molecular weight extraction kits) and two ONT library preparation protocols (ligation and the rapid barcoding sequencing kit) using five well-characterized strains representing diverse foodborne pathogens. High-quality, closed bacterial genomes were obtained from all combinations of extraction and sequencing kits. However, variations in assembly length and genome completeness were observed, indicating the need for further optimization. In silico analyses demonstrated that Q20+ nanopore sequencing chemistry accurately identified species, genotype, and virulence factors, with comparable results to Illumina sequencing. Phylogenomic clustering showed that ONT assemblies clustered with reference genomes, though some indels and SNP differences were observed, likely due to sequencing and analysis methodologies rather than inherent genetic variation. Additionally, the study evaluated the impact of a change in the sampling rates from 4 kHz (260 bases pair second) to 5 kHz (400 bases pair second), finding no significant difference in sequencing accuracy. This evaluation workflow offers a framework for evaluating novel technologies for use in surveillance and foodborne outbreak investigations. Overall, the evaluation demonstrated the potential of ONT Q20+ nanopore sequencing chemistry to assist in identifying the correct strain during outbreak investigations. However, further research, validation studies, and optimization efforts are needed to address the observed limitations and fully realize the technology’s potential for improving public health outcomes and enabling more efficient responses to foodborne disease threats.

1. Introduction

Leafy greens are responsible for nearly half of the produce-related Shiga toxin-producing Escherichia coli (STEC) outbreaks in the United States, and recent investigations have implicated agricultural water as a potential source [1]. Currently, the Food and Drug Administration (FDA) outbreak detection protocols for STEC require extensive analysis time (2–4 weeks) [2]. During outbreak investigations, finding the strain that matches the outbreak strain can take several days after isolating each strain from the potential food sources. This delay is crucial, considering that leafy greens have a limited shelf life. During outbreak events and periods of uncertainty, methods are needed to safely preserve these products to prevent unnecessary waste. Therefore, there is a need for methods that can assist in rapidly identifying the correct strain and producing a preliminary sequencing analysis during such investigations, potentially reducing the time required to find the outbreak strain.
Currently, the FDA utilizes Illumina instruments for whole genome sequencing (WGS) analysis of isolated strains for surveillance [3,4,5]. While short-read Illumina MiSeq sequencing technology is highly accurate, it struggles with assembling highly repetitive regions that can span several hundred base pairs [6,7,8,9,10,11]. As a result, the final assembled genome of an outbreak strain typically consists of numerous contigs (e.g., approximately 200 contigs for STECs) and often lacks critical information such as plasmid presence and the location of important genes, including antimicrobial resistance genes (AMR) [10,12,13,14,15]. An alternative approach involves using Oxford Nanopore Technologies (ONT) sequencing, a technology that allows for determining a DNA or RNA sequence by reading DNA or RNA through a biological nanopore and measuring changes in an electrical current that translates into a base call. ONT produces long reads that can identify taxa while sequencing. This method has shown promise in outbreak detection [8,9], but its lower raw read accuracy previously made it unsuitable as a stand-alone technique for certain downstream analyses, such as SNP analysis during outbreak investigations [3,4,16,17,18]. Accurate SNP variant detection is crucial in outbreak investigations, as some strains can differ by only a few SNPs (around 1–10) [3,4,18,19]. However, in 2023, ONT released a new chemistry (Q20+), achieving ≥99% raw read accuracy and high sequencing yield [20,21,22,23,24], making it a potential candidate for improved and faster outbreak response.
Prior to implementation of ONT sequencing as an alternative as an enhanced and faster tool for outbreak response, it is crucial to ensure that the quality obtained is on par with the current gold standard, Illumina sequencing. To achieve this, a pilot study of ONT’s new chemistry, including the V14 DNA library preparation kits and new R10.4.1 flow cell, is necessary to select the optimal combination to perform a more rigorous single laboratory validation (SLV). According to the FDA’s Guidelines for the Validation of Analytical Methods Using Nucleic Acid Sequenced-Based Technologies [25], several important considerations must be taken into account when conducting a verification study for a new change in next-generation sequencing (NGS) chemistry such as the Q20+ from ONT. These considerations include (1) the use of a well-characterized isolate panel for primary NGS/WGS data collection, (2) reproducibility (precision between runs), and (3) repeatability (precision within a run). In this pilot study, only the first consideration will be evaluated, and it will be conducted using a set of reference genomes. Among the parameters to evaluate are (a) average depth of genome coverage, (b) mean read length, (c) sequence length distribution, (d) assembly length, (e) number of contigs, (f) N50, and (g) percentage of recovered open reading frames (ORF) and core genes. Subsequently, in silico tests will be conducted to assess other aspects, such as correct species identification (accuracy, specificity, and sensitivity), genotyping methods [serotyping, AMR, virulence, toxin, multi-locus sequence typing (MLST)], and phylogenomic clustering methods measuring specificity and sensitivity for cluster assignment. A wgMLST approach using the known genome reference of the tested strains as a reference will be employed to determine specificity and sensitivity for determining variants (single nucleotide polymorphisms (SNPS) and/or indels) and the accuracy of the entire tree or specific splits leading to an outbreak lineage. The entire workflow is depicted in Figure 1.
ONT has two main DNA library preparation kits for whole genome sequencing of bacteria: (1) the Ligation Sequencing Kit (SQK-LSK114) and (2) the Rapid Barcoding Sequencing Kit (SQK-RBK11). They differ in the preparation time and steps required to obtain a library ready for sequencing. The RBK kit is faster and less labor-intensive than the LSK kit, albeit producing lower sequencing output. The sequencing performance and output can be highly impacted by the quality of the DNA used for the sequencing. Therefore, we chose to additionally evaluate the performance of these two library preparation kits with two DNA extraction kits that produced DNA with different specifications: (1) the faster and automatized Maxwell RSC Cultured Cell DNA kit (high DNA concentration output but sheared DNA) and (2) the manual Monarch High Molecular Weight (HMW) DNA Extraction Kit (DNA concentration high and with more integrity—longer fragments). For this pilot study, we selected for sequencing five well known and characterized foodborne and healthcare-associated bacterial pathogen strains that pose significant public health threats, belonging to five different species from our collection: Salmonella enterica subsp. enterica, Vibrio parahemolyticus, Shigella sonnei, Escherichia coli, and Klebsiella pneumoniae (Table 1). These organisms are closely monitored by health services due to their roles in outbreaks, potential for severe illness, and emerging antibiotic resistance. The DNA was extracted from overnight cultures with the two different kits and sequenced using the two ONT library preparation kits separately, for a total of four runs, as shown in Table 2 and Supplementary Figure S1.

2. Results

Preliminary testing of the Q20+ chemistry was done to establish which sequencing kit and DNA combination could be used for easily and completely closing bacterial genomes for later validation efforts. As shown in Table 2, each combination resulted in different total output and total read numbers. The highest output both in read numbers and sequence was achieved with the LSK114/Maxwell combination (10.65 Gb and 1.76 M reads). The lowest was obtained with the RBK114/NEB combination (3.28 Gb and 305k reads). The higher N50 length reads per strain per run were observed on the runs using the DNA extracted with the NEB HMW kit with any of the two DNA library combinations (Supplementary Figure S2). The higher number of reads per strain per run were observed on the runs using the Maxwell/LSK114 combination. This partially explains why the RBK114/NEB run had the lowest output.

2.1. Genome Sequencing QC

The data generated from the four runs were compared for assembly statistics: (a) assembly length (bp), (b) average depth of genome coverage, (c) genome completeness, (d) number of contigs, and (e) % ORF recovered. Table 3 shows the assembly length for each strain per run and the average depth coverage for each genome per run. The assembly length per strain per run showed great variation. For the Salmonella strain, the assembled genome length varied from 2 kb less to 5 kb more compared to its reference genome; for the Vibrio parahemolyticus, it varied from 5 kb to 38 kb more compared to the reference genome; for the Shigella sonnei it varied from 17 kb less to 21 kb more compared to its reference genome; for the E. coli O157:H7, it varied from 2 kb less to 22 kb more compared to its reference genome; and for the Klebsiella pneumoniae, the variation was less pronounced, from 0.2 kb to 3.8 kb more compared to its reference genome. These variations do not appear to be linked to the sequencing depth, since, for example, Klebsiella pneumoniae showed the higher variation for the highest coverage (>300×).
Regarding genome completeness, most genomes were complete except for CFSAN076620 (E. coli O157:H7, run RBK114-NEB and LSK114-NEB) and CFSAN086181 (Klebsiella pneumoniae, run RBK114-NEB) (Table 4). Most of the ORFs (>99%) were recovered by ONT Q20+ in all of the combinations compared to the reference genome for each strain (Table 5). Based on the analyzed genome and sequencing combinations, there were still issues with indels, and the recovery of some ORFs was not possible. The results were not identical and showed inconsistencies between runs.

2.2. In Silico Analyses

The data generated from the four runs were compared for diverse in silico tests or assays: (a) correct species identification (accuracy, specificity, and sensitivity), (b) genotyping methods (serotyping, AMR, virulence, toxin, MLST), and (c) phylogenomic clustering methods measuring the specificity and sensitivity for cluster assignment.
(a)
Correct species identification (accuracy, specificity, and sensitivity). Two different methods for species identification were used. The initial identification was performed using the WIMP workflow, and each species was identified correctly. This initial assessment was confirmed using a different taxonomic tool (Kraken 2) using the final assemblies for each sample. The assemblies for each strain from each run were run through it, and each species was identified correctly (100% accuracy, specificity, and sensitivity).
(b)
Genotyping methods (serotyping, AMR, virulence, toxin, MLST). Regarding the serotyping, only two serotyping schemes were available for the five different species included in this pilot study (Salmonella and E. coli). There was a 100% agreement with the serotype predictions, as all Salmonella and E. coli ONT assemblies were identified as serotype Bareilly and O157:H7, respectively. Similarly, in silico sequence type (ST) determination yielded identical results for ONT assembly (ST-909 for CFSAN000189, ST-635 for CFSAN123154, ST-152 for CFSAN030807, ST-11 for CFSAN076620, and ST-629 for CFSAN086181), except for one assembly (CFSAN000189) that showed an incomplete MLST profile due to an indel in the sucA gene (NEB-LSK114). All ONT assemblies for each strain agreed 100% for the presence of specific antimicrobial resistance (AMR) genes compared to the reference genome for each strain. CFSAN000189 carried a single copy of the aac(6′)-Iaa gene, CFSAN123154 carried a single AMR gene (blaCARB-36), CFSAN076620 carried two copies of the catA1 gene, and CFSAN086181 carried five genes (blaSHV-110, blaSHV-81, fosA6, OqxB, and OqxA). An issue arose with the copy number for the AMR genes for strain CFSAN030807, which carried four AMR genes [aph(6)-Id, aph(3″)-Ib, sul2, and tet(A)]. However, in the assemblies for runs M-LSK114 and M-RBK114, the plasmid carrying the AMR genes was double in size compared to the original plasmid, likely due to an assembly issue (we have observed this same issue in many Q20+ ONT assemblies), resulting in the same sequence being concatenated twice.
For this pilot study, in silico analyses were conducted on ONT assemblies of the five strains to determine the presence of virulence and toxin genes. The genes reported for each species in the virulence factors database (VFDB) task template included in the Ridom Seqsphere software v9.0.8 were utilized for these analyses, except for E. coli, where Virulence Finder at DTU was used instead. The reference genome of Salmonella enterica subsp. enterica (CFSAN000189) Bareilly contained 105 virulence genes, all of which were consistently identified in ONT assemblies. Similarly, the Vibrio parahemolyticus (CFSAN123154) strain, characterized as tdh−/trh+, contained 42 virulence genes in its reference genome, all of which were also found in every ONT assembly. Additionally, the Shigella sonnei (CFSAN030807) strain’s reference genome contained 10 virulence genes, all of which were consistently identified in ONT assemblies. This strain has lost a larger plasmid that carried 42 additional virulence genes compared to the current genome available at NCBI. In the case of E. coli (CFSAN076620), the reference genome contained 20 relevant virulence genes, which were consistently detected across different ONT combinations, with only one gene (toxB) missing in certain ONT assemblies, specifically NEB-RBK114 and NEB-LSK114. Finally, the reference genome of Klebsiella pneumoniae (CFSAN086181) harbored 52 virulence genes, all of which were present in every ONT assembly.
(c)
Phylogenomic clustering methods measuring the specificity and sensitivity for cluster assignment. For this last step, a wgMLST approach was used to determine specificity (using as a reference genome the known genome sequence of the tested strains) and sensitivity for determining variants (SNPS and/or indels), accuracy of entire tree, or specific splits (e.g., leading to an outbreak lineage).

2.3. wgMLST Analyses for CFSAN000189 (Salmonella Bareilly)

For this step, two different wgMLST analyses were performed: (1) against a collection of closely related Salmonella Bareilly genomes and (2) against the reference genome for that same strain (obtained by PacBio or by a hybrid assembly of ONT and Illumina data). Total genes used in the wgMLST were 4312 loci. Of those, 3991 loci (93%) were identical among all samples (including the PacBio and hybrid genome). The ONT assemblies obtained with the Q20+ chemistry (V14) by either combination clustered with the corresponding genome in a SNP NJ tree after a wgMLST analysis (Figure 2A), albeit with at least 100 to 124 SNPs and 30–39 indels, compared to the reference genome (Figure 2B and Table 6). The SNP differences matrix is shown in Table 6. We must add that this strain was sequenced using the same initial bacterial culture; thus, the DNA must be the same in terms of methylations. The complete NJ SNP tree can be found on Supplementary Figure S3. Additionally, the Illumina sequences [obtained using a MiSeq (M)] assembled were also included in the wgMLST analysis and were indistinguishable from the PacBio or ONT hybrid assemblies.

2.4. wgMLST Analyses for CFSAN123154 (Vibrio parahemolyticus)

Similar to the approach used for Salmonella, two different wgMLST analyses were performed for the Vibrio parahemolyticus strain: (1) against a collection of closely related Vibrio parahaemolyticus genomes and (2) against the reference genome for that same strain (obtained by PacBio). The total genes used in the wgMLST were 4440 loci. Of those, 4078 loci (92%) were identical among all samples (same DNA prepared with different library kit and run in different flow cells). The ONT assemblies obtained with the Q20+ chemistry (V14) by either combination clustered with the corresponding genome in an SNP NJ tree after a wgMLST analysis (Figure 3A), albeit with at least 100 to 133 SNPs and 53 to 81 indels, compared to the reference genome (Figure 3B and Table 7). The complete NJ SNP tree can be found on Supplementary Figure S4. The SNP differences matrix is shown in Table 7.

2.5. wgMLST Analyses for CFSAN030807 (Shigella sonnei)

Similar to the approach used for Salmonella, two different wgMLST analyses were performed for the Shigella sonnei strain: (1) against a collection of closely related Shigella sonnei genomes (ST-152) and (2) against the reference genome for that same strain (obtained by PacBio). Total genes used in the wgMLST were 3699 loci. Of those, 3657 loci (99%) were identical among all samples (same DNA prepared with different library kit and run in different flow cells). The ONT assemblies obtained with the Q20+ chemistry (V14) by either combination were clustered with the corresponding genome in a SNP NJ tree after a wgMLST analysis (Figure 4A), with at least 0 to 2 SNPs and 1 to 13 indels, compared to the reference genome (Figure 4B and Table 8). The SNP differences matrix is shown in Table 8. We must add that this strain was sequenced using the same initial bacterial culture, so the DNA must be the same in terms of methylations. The complete NJ SNP tree can be found in Supplementary Figure S5.

2.6. wgMLST Analyses for CFSAN076620 (E. coli O157:H7)

Similar to the approach used for Salmonella, two different wgMLST analyses were performed for the E. coli strain: (1) against a collection of closely related E. coli O157:H7 genomes (ST-11) and (2) against the reference genome for that same strain (obtained by PacBio). Total genes used in the wgMLST were 4639 loci. Of those, 4500 loci (97%) were identical among all samples (same DNA prepared with different library kits and run in different flow cells). The ONT assemblies obtained with the Q20+ chemistry (V14) by either combination clustered with the corresponding genome in a SNP NJ tree after a wgMLST analysis (Figure 5A), albeit with at least 3 to 28 SNPs and 32 to 74 indels, were compared to the reference genome (Figure 5B and Table 9). The SNP differences matrix is shown in Table 9. We must add that this strain was sequenced using the same initial bacterial culture, so the DNA must be the same in terms of methylations.

2.7. wgMLST Analyses for CFSAN086181 (Klebsiella pneumoniae)

Similar to the approach used for Salmonella, two different wgMLST analyses were performed for the Klebsiella pneumoniae strain: (1) against a collection of closely related Klebsiella pneumoniae genomes and (2) against the reference genome for that same strain (obtained by PacBio). Total genes used in the wgMLST were 4841 loci. Of those, 4778 loci (98.7%) were identical among all samples (same DNA prepared with different library kit and run in different flow cells). The ONT assemblies obtained with the Q20+ chemistry (V14) by either combination clustered with the corresponding genome in a SNP NJ tree after a wgMLST analysis (Figure 6A), with one to nine SNPs and 10 to 24 indels compared to the reference genome (Figure 6B and Table 4). The SNP differences matrix is shown in Table 10. Note that this strain was sequenced using the same initial bacterial culture, so the DNA should be identical in terms of methylations. The complete NJ SNP tree can be found in Supplementary Figure S6.

2.8. Testing the New Software Update (MinKNOW 23.04.5) for the Same Samples Using the RBK-114 Kit

The new software update (MinKNOW 23.04.5) was tested using the RBK-114 kit on the same samples. During the evaluation of the new Q20+ chemistry, ONT released an update for the MinKNOW software (v23.04.5), which incorporated a new sampling rate (5 kHz) and basecaller model ([email protected]). ONT claimed that there were no differences in the accuracy of the SUP 400 bps (5 kHz) compared to the previous SUP at 260 bps (4 kHz). A test using the same five DNA strains and the rapid barcoding kit (RBK-114) was performed. The output was very similar to the observed RBK runs (741.66k reads and 3.9 Gb, flow cell number FAW81132). The procedure for data analysis remained consistent, employing two different wgMLST analyses for each strain: (1) against a collection of closely related genomes and (2) against the reference genome for the same strain. Every strain clustered in the same cluster for each individual species, as observed with the SUP 260 bps. Regarding indels and SNPs,
-
CFSAN000189 had 32 SNPs and 56 indels compared to its reference genome,
-
CFSAN123154 had 77 SNPs and 78 indels compared to its reference genome,
-
CFSAN030807 had 0 SNPs and 5 indels compared to its reference genome,
-
CFSAN076620 had 3 SNPs and 27 indels compared to its reference genome, and
-
CFSAN086181 had 2 SNPs and 7 indels compared to its reference genome.

3. Discussion

This study was conducted to address the limitations of current FDA pathogen sequencing methods, which are time consuming and require expensive equipment. One possible solution is the use of less expensive sequencers, which could increase access to this type of technology for many potential users. Expanding access to this type of technology is essential to enhance the effectiveness and speed of foodborne outbreak responses. This research aimed to evaluate whether Oxford Nanopore Technologies (ONT) MinION with Q20+ chemistry can offer rapid and accurate identification of foodborne pathogens, including SNP differences, serotypes, and antimicrobial resistance genes. The goal was to assess if ONT Q20+ technology could provide near real-time pathogen identification, potentially overcoming the drawbacks of current methodologies. The study aimed to bridge advanced genomic technologies with regulatory science, enabling more effective responses to foodborne disease threats and enhancing public health outcomes.
To achieve this, the study evaluated ONT Q20+ chemistry using five well-characterized strains representing diverse foodborne pathogen traits. Various DNA extraction (Maxwell and NEB HMW) and library preparation methods (RBK-114 and LSK-114) were compared to optimize sequencing outcomes. Any combination proved sufficient to generate high-quality bacterial genomes using the Flye assembler as observed by numerous authors under similar conditions [20,21,22]. However, the ligation kit (LSK-114) produced a higher output and longer reads when used in combination with the NEB HMW DNA extraction kit, as observed previously by other authors [26,27]. If you want to assemble very complex bacterial genomes, the combination of high molecular weight DNA extraction (less sheared DNA) and the ligation sequencing kit (SQK-LSK114) is recommended as observed for STECs [28,29]. For further validation studies, the combination of the rapid library sequencing kit (RBK-114) and the automated DNA extraction kit (Maxwell RSC Cultured Cell DNA kit) proved to be optimal and preferred because of speed and reduced errors during DNA extraction and sequencing preparations resulting in less laboratory discrepancies. Variations in assembly length and genome completeness were observed across different combinations of extraction and sequencing kits. The differences observed in genome completeness and assembly length for the STEC and Klebsiella pneumoniae are likely due to variations in the downstream assembly process rather than the DNA extraction or library preparation methods. Given that both methods consistently achieved sufficient coverage (above 30X), we do not expect the extraction or preparation methods to significantly impact genome quality, as the primary difference between the methods relates to total read yield rather than assembly accuracy. Most genomes achieved high completeness (>99% ORF recovery) with ONT Q20+, as has been observed previously [22,23,24], but some variations in assembly length and missing ORFs were noted, suggesting the need for further optimization. In addition, duplication of small plasmids was observed in some cases when using the Flye assembler, as observed by others [21,30,31,32], specifically for strain CFSAN030807, where the expected small plasmid of 8401 bp carrying four AMR genes was 16,802 bp in size for the assemblies of M—RBK114 and M—LSK114. Therefore, the reason the plasmid is double in size seems random, or perhaps the DNA extracted with that specific kit created smaller fragments for that plasmid that were randomly misassembled due to some minor SNP differences.
Regarding the in silico analyses, the Q20+ nanopore sequencing chemistry demonstrated 100% for accurate species identification, genotyping (including serotyping and AMR gene detection), and virulence factor identification comparable to Illumina sequencing. Phylogenomic clustering methods (wgMLST) showed that ONT assemblies clustered with reference genomes, although some indels and SNP differences were observed with some bacterial species and less pronounced with others. Supplementing with Illumina data helped resolve these differences. Other authors have similar results for other enterobacterial species (Citrobacter freundii, Klebsiella pneumoniae, Enterobacter hormaechei, Enterobacter cloacae, and Escherichia coli) and noticed that there was still a need to supplement with Illumina data in order to correct those indels or SNPs [21]. Despite these variations, the overall clustering and similarity to reference genomes highlight the robustness of nanopore sequencing results for assisting in detecting the correct strain during outbreak investigations. While SNP variation is important, genomes within an outbreak may differ by a few SNPs due to natural mutation rates during strain culturing. Therefore, integrating genetic data with epidemiological context is essential for accurately interpreting phylogenetic relationships and assessing potential outbreak inclusion, ultimately enhancing our ability to draw informed conclusions. The SNPs observed were randomly distributed in the genome of some of the analyzed strains. This addition clarifies that any observed SNP differences in the phylogenetic analyses likely stemmed from sequencing and analysis processes rather than genetic variations in the sampled bacteria. It is important to note that these analyses were conducted using DNA from the same initial bacterial culture, and the impact of methylations on the basecallers should have been the same for each strain and should not reflect on the SNPs observed for each run for that specific strain (Supplementary Figure S6).
The assembled genomes from each combination were compared against their reference genome annotated coding sequences (CDS) using a highly conservative requirement for a 100% match, as described by Sanderson et al. (2024) [22]. Both indels and SNP variants were identified using this process. Regulatory actions demand a high level of certainty to identify an outbreak strain. Even though we performed a single run per combination (except for the Maxell RBK-114 combination), we believe that it is equivalent to multiple replicates of the same sample. Although only a single run was performed for each combination (except for the Maxell RBK-114 combination), this is considered equivalent to multiple replicates of the same sample. Since the same bacterial culture and chemistry were used, with only variations in library preparation kits and procedures, the final sequence data quality should remain unaffected. As mentioned by Sanderson et al. 2024 [22], factors such as pore number, operator skill, and library type can influence read output and yield. Additionally, while only five strains from different species were analyzed, Sanderson et al. (2024) cautioned that these findings may not be broadly applicable across all strains or species. Most studies evaluating the newer ONT chemistry focus on a single species or a single sequencing run, limiting the scope of the technology’s assessment [20,23,33,34]. In contrast, this study expands the evaluation by testing five different bacterial species and performing five sequencing run replicates from the same initial culture. This approach enhances the analytical robustness of the study, allowing for a more comprehensive evaluation. Additionally, it assesses sequencing errors to determine whether they consistently occur at the same genomic positions—potentially linked to methylation patterns—or if they are randomly distributed as shown in this study. Our results also demonstrated that assembly accuracy varied by species/strain compared to the reference (Table 6, Table 7, Table 8, Table 9 and Table 10). A key innovation of this study is that it demonstrates the potential of ONT newer chemistry to deliver more accurate sequences than the previous ONT chemistry (R9.4.1), faster and at a lower initial cost, making it a more efficient tool for bacterial screening during outbreak investigations.
During the initial evaluation of the Q20+ chemistry, the study coincided with a major update to the basecaller. Using the RBK-114 kit, the same five DNA strains were processed with both the new and old basecaller models. ONT claimed no difference in accuracy between the SUP 400 bps (5 kHz) and the previous SUP 260 bps (4 kHz) models, and our findings largely supported this. Additionally, downstream analyses showed that clustering for all strains was consistent with the results obtained using the SUP 260 bps model, indicating that the new basecaller model did not affect overall clustering or comparative accuracy in detecting SNPs and indels. Overall, testing confirmed that the update to MinKNOW 23.04.5 and its new basecaller model maintains accuracy and performs comparably to the previous version, validating ONT’s claims. This suggests that users can confidently transition to the new software update without concerns over data quality loss. However, it should be noted that ONT has made several basecaller changes since the start of the evaluation, which could raise concerns for both current and potential new users.

4. Materials and Methods

4.1. Bacterial Strains and Media

The bacterial strains used in this study belonged to 5 different species associated with foodborne outbreak or cases collected by the Center for Food Safety and Applied Nutrition (CFSAN) as part of our Genome Trakr surveillance program (Table 1). These 5 strains were selected to represent various species, antimicrobial resistance (AMR) genes, common foodborne pathogen, diverse % GC content (medium and high), SNP recovery as part of the same outbreak, toxin gene detection, and diverse plasmid content. Bacterial strains were cultured in brain heart infusion (BHI) broth (Thermo Fisher Scientific, Waltham, MA, USA) and stored in BHI broth with 20% glycerol stocks at −80 °C.

4.2. DNA Extraction and Quality and Quantification Measurement

Genomic DNA from each strain was extracted using either the Maxwell RSC Cultured Cell DNA Kit with a Maxwell RSC Instrument (Promega Corporation, Madison, WI, USA) according to the manufacturer’s instructions for Gram-negative bacteria with additional RNase treatment or the Monarch HMW DNA Extraction kit (New England Biolabs, Ipswich, MA, USA). DNA concentration was determined by a Qubit 4 Fluorometer (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. The DNA integrity was determined using a Femto Pulse System (Agilent, Santa Clara, CA, USA).

4.3. Whole Genome Sequencing and Assembly Using Long Reads

DNA extracted from each strain was sequenced using the GridION Sequencing Device Mk1 (Oxford Nanopore Technologies, Oxford, UK) in different conformations (Table 2). The sequencing libraries were prepared using different protocols: (1) Rapid Barcoding Kit 24 V14 (SQK-RBK114.24) or (2) Ligation Sequencing Kit v14 (SQK-LSK114) with Native Barcoding Kit (EXP-NBD114). They were then run in FLO-MIN114 (R10.4.1) flow cells, according to the manufacturer’s instructions, for 48 h (Oxford Nanopore Technologies). The sequencing run was live base called using different MinKNOW software versions (see Table 2) using the super-accurate basecalling model (SUP, basecall_model—[email protected]). The reads < 2000 bp and quality scores of <10 were discarded for downstream analysis using Fitlong v0.2.1 (https://github.com/rrwick/Filtlong, accessed on 30 October 2024) with default parameters. The genomes for each strain were obtained by de novo assembly using the Flye program v2.9.2 [35], using the following parameters: --nano-hq, --read-error 0.03, and -i 4. Flow cell FAW81132 was run using a different sampling rate (5 kHz) and basecalling method: [email protected].
The short read whole genome sequences for these strains were generated by Illumina MiSeq sequencing with the MiSeq V3 kit using 2 × 250 bp paired-end chemistry (Illumina, San Diego, CA, USA), according to the manufacturer’s instructions, at 100X coverage. The libraries were constructed using 100 ng of genomic DNA using the Illumina® DNA Prep (M) Tagmentation (Illumina, San Diego, CA, USA), according to the manufacturer’s instructions. Reads were trimmed with Trimmomatic v0.36 [36] using default parameters.

4.4. Hybrid Assembly

The complete genome for each strain from run FAW25027 was obtained by a de novo hybrid assembly using both nanopore and MiSeq data for that strain with Unicycler v0.5.0 [37]. Unicycler assembled the chromosome and plasmids as circular closed and oriented the chromosome to start at the dnaA gene.

4.5. Species Identification

The initial species identification was performed using the reads as they were being produced by the sequencing device for every single sample and analyzed using the ONT “What’s in my pot” (WIMP) workflow contained in the EPI2ME cloud service (Oxford Nanopore Technologies) after 5 h. The final taxonomical classification and species confirmation were performed using Kraken 2 [38] implemented in the GalaxyTrakr website [39] using the default database.

4.6. In Silico MLST and Serotyping

The initial analysis and identification of the strains were performed using an in silico MLST approach using Ridom SeqSphere+ v9.0.8 (Ridom, Münster, Germany) using their MLST task template for each individual species as follows: for Salmonella enterica (https://enterobase.warwick.ac.uk/species/index/senterica, all accessed on 30 October 2024), Vibrio parahemolyticus (https://pubmlst.org/organisms/vibrio-parahaemolyticus), Shigella sonnei (https://enterobase.warwick.ac.uk/species/index/ecoli), Escherichia coli (https://enterobase.warwick.ac.uk/species/index/ecoli), and Klebsiella pneumoniae (https://bigsdb.pasteur.fr/klebsiella/) (Salmonella and E. coli). Only two serotyping schemes were available for the 5 different species included in this pilot study (Salmonella and E. coli). For Salmonella enterica we used SeqSero2 v1.1.0 (http://www.denglab.info/SeqSero2), and for E. coli we used SerotypeFinder 2.0 (https://cge.food.dtu.dk/services/SerotypeFinder/) [40].

4.7. In Silico AMR Gene Detection

The AMR gene detection was performed using ResFinder v4.5.0 (http://genepi.food.dtu.dk/resfinder) [41].

4.8. In Silico Identification of Virulence Genes

The de novo assemblies were batch screened for virulence genes using the Ridom SeqSphere+ task template for the virulence genes reported for each species in the virulence factors database (VFDB—http://www.mgc.ac.cn/VFs/main.htm), except for E. coli, where Virulence Finder at DTU was used instead.

4.9. Phylogenetic Relationship of the Strains Determined by wgMLST Analysis

The phylogenetic relationship of the strains was assessed by a whole genome multilocus sequence typing (wgMLST) analysis using Ridom SeqSphere+ v9.0.8. A separate wgMLST was created for each individual species using the genome available at NCBI or obtained in house for that same strain as reference (Table 1). Genes that were repeated in more than one copy in any of the two genomes were removed from the analysis as failed genes. A task template was then created that contains both core and accessory genes for each reference strain for any future testing. Each individual locus (core or accessory genes) from the reference strain was assigned allele number 1. The assemblies for each individual strain closed genome in this study were queried against the task template for each individual species/strain. If the locus was found and was different from the reference genome or any other queried genome already in the database, a new number was assigned to that locus and so on. After eliminating any loci that were missing or found containing indels (showed as failed call in the software) from the genome of any strain used in our analyses, we performed the wgMLST analysis. These remaining loci were considered the core genome shared by the analyzed strains. We used Nei’s DNA distance method [42] to calculate the matrix of genetic distance, considering only the number of same/different alleles in the core genes. A neighbor-joining (NJ) tree was built using pairwise ignoring missing values and the appropriate genetic distances after the wgMLST analysis. wgMLST uses the alleles number of each locus to determine the genetic distance and build the phylogenetic tree. The use of allele numbers reduces the influence of recombination in the dataset studied and allows for fast clustering determination of genomes.

4.10. Nucleotide Sequence Accession Numbers

The raw ONT data for each individual sample were deposited in GenBank under BioProject accession number PRJNA1134675. Each sample SRA number matching its corresponding sample can be found in Supplementary Table S1.

5. Conclusions

The evaluation of the Q20+ nanopore sequencing chemistry in this single laboratory evaluation study demonstrates its potential for assisting in finding the correct strain during outbreak investigations. Standardizing the extraction and sequencing protocols is crucial for optimizing workflow and ensuring reproducibility and accuracy across different strains and outbreak scenarios. While the study encountered challenges in achieving consistent and reliable results with ONT Q20+ chemistry, the findings provide valuable insights into the technology’s potential for enhancing foodborne pathogen detection and surveillance. Continued research and optimization efforts are necessary to address the observed limitations and fully realize the impact of nanopore sequencing on public health and regulatory science. Further verification studies and optimizations are essential to integrate nanopore sequencing into routine outbreak surveillance and response protocols. Overcoming current limitations will be crucial for incorporating these technologies into routine foodborne pathogen surveillance, ultimately leading to improved public health outcomes and more agile responses to foodborne disease threats. The accuracy of SNPs in the assemblies compared to Illumina or PacBio assemblies remains an issue. This is being addressed in future iterations of the ONT basecaller, but the improvements need to be properly assessed and verified.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/ijms252211877/s1.

Author Contributions

Conceived and designed the experiments: N.G.-E. and M.H.; Analyzed the data: N.G.-E. and M.H.; Contributed reagents/materials/analysis tools: N.G.-E., M.H., J.H.J. and S.M.T.; Wrote the paper: N.G.-E.; Revision and final draft: M.H., J.H.J., S.M.T. and N.G.-E. All authors have read and agreed to the published version of the manuscript.

Funding

This study was supported by funding from the FDA Foods Program Intramural Funds. Jae Hee acknowledges a Research Fellowship Program (ORISE) for the Center for Food Safety and Applied Nutrition administered by the Oak Ridge Associated Universities through a contract with the U.S. Food and Drug Administration.

Data Availability Statement

Data is contained within the article and Supplementary Materials.

Conflicts of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References

  1. BoBottichio, L.; Keaton, A.; Thomas, D.; Fulton, T.; Tiffany, A.; Frick, A.; Mattioli, M.; Kahler, A.; Murphy, J.; Otto, M.; et al. Shiga toxin-producing Escherichia coli infections associated with romaine lettuce-united states, 2018. Clin. Infect. Dis. 2020, 71, e323–e330. [Google Scholar] [CrossRef] [PubMed]
  2. Feng, P.; Weagant, S.D.; Jinneman, K. BAM Chapter 4A: Diarrheagenic Escherichia coli. Available online: https://www.fda.gov/food/laboratory-methods-food/bam-chapter-4a-diarrheagenic-escherichia-coli (accessed on 30 October 2024).
  3. Timme, R.E.; Strain, E.; Baugher, J.D.; Davis, S.; Gonzalez-Escalona, N.; Sanchez Leon, M.; Allard, M.W.; Brown, E.W.; Tallent, S.; Rand, H. Phylogenomic pipeline validation for foodborne pathogen disease surveillance. J. Clin. Microbiol. 2019, 57, e01816-18. [Google Scholar] [CrossRef] [PubMed]
  4. Allard, M.W.; Bell, R.; Ferreira, C.M.; Gonzalez-Escalona, N.; Hoffmann, M.; Muruvanda, T.; Ottesen, A.; Ramachandran, P.; Reed, E.; Sharma, S.; et al. Genomics of foodborne pathogens for microbial food safety. Curr. Opin. Biotechnol. 2018, 49, 224–229. [Google Scholar] [CrossRef] [PubMed]
  5. Timme, R.E.; Pettengill, J.B.; Allard, M.W.; Strain, E.; Barrangou, R.; Wehnes, C.; Van Kessel, J.S.; Karns, J.S.; Musser, S.M.; Brown, E.W. Phylogenetic diversity of the enteric pathogen Salmonella enterica subsp. enterica inferred from genome-wide reference-free SNP characters. Genome Biol. Evol. 2013, 5, 2109–2123. [Google Scholar] [CrossRef] [PubMed]
  6. Bertrand, D.; Shaw, J.; Kalathiyappan, M.; Ng, A.H.Q.; Kumar, M.S.; Li, C.; Dvornicic, M.; Soldo, J.P.; Koh, J.Y.; Tong, C.; et al. Hybrid metagenomic assembly enables high-resolution analysis of resistance determinants and mobile elements in human microbiomes. Nat. Biotechnol. 2019, 37, 937–944. [Google Scholar] [CrossRef]
  7. Moss, E.L.; Maghini, D.G.; Bhatt, A.S. Complete, closed bacterial genomes from microbiomes using nanopore sequencing. Nat. Biotechnol. 2020, 38, 701–707. [Google Scholar] [CrossRef] [PubMed] [PubMed Central]
  8. Maguire, M.; Ramachandran, P.; Tallent, S.; Mammel, M.K.; Brown, E.W.; Allard, M.W.; Musser, S.M.; Gonzalez-Escalona, N. Precision metagenomics sequencing for food safety: Hybrid assembly of Shiga toxin-producing Escherichia coli in enriched agricultural water. Front. Microbiol. 2023, 14, 1221668. [Google Scholar] [CrossRef] [PubMed]
  9. Maguire, M.; Kase, J.A.; Roberson, D.; Muruvanda, T.; Brown, E.W.; Allard, M.; Musser, S.M.; Gonzalez-Escalona, N. Precision long-read metagenomics sequencing for food safety by detection and assembly of Shiga toxin-producing Escherichia coli in irrigation water. PLoS ONE 2021, 16, e0245172. [Google Scholar] [CrossRef] [PubMed]
  10. Gonzalez-Escalona, N.; Kase, J.A. Virulence gene profiles and phylogeny of Shiga toxin-positive Escherichia coli strains isolated from FDA regulated foods during 2010–2017. PLoS ONE 2019, 14, e0214620. [Google Scholar] [CrossRef] [PubMed]
  11. Wagner, D.D.; Carleton, H.A.; Trees, E.; Katz, L.S. Evaluating whole-genome sequencing quality metrics for enteric pathogen outbreaks. PeerJ 2021, 9, e12446. [Google Scholar] [CrossRef] [PubMed]
  12. Gonzalez-Escalona, N.; Toro, M.; Rump, L.V.; Cao, G.; Nagaraja, T.G.; Meng, J. Virulence gene profiles and clonal relationships of Escherichia coli oO26:H11 isolates from feedlot cattle as determined by whole-genome sequencing. Appl. Environ. Microbiol 2016, 82, 3900–3912. [Google Scholar] [CrossRef] [PubMed]
  13. Haendiges, J.; Jinneman, K.; Gonzalez-Escalona, N. Choice of library preparation affects sequence quality, genome assembly, and precise in silico prediction of virulence genes in shiga toxin-producing Escherichia coli. PLoS ONE 2021, 16, e0242294. [Google Scholar] [CrossRef] [PubMed]
  14. Rusconi, B.; Sanjar, F.; Koenig, S.S.; Mammel, M.K.; Tarr, P.I.; Eppinger, M. Whole Genome Sequencing for Genomics-Guided Investigations of Escherichia coli O157:H7 Outbreaks. Front. Microbiol. 2016, 7, 985. [Google Scholar] [CrossRef] [PubMed]
  15. Eppinger, M.; Daugherty, S.; Agrawal, S.; Galens, K.; Sengamalay, N.; Sadzewicz, L.; Tallon, L.; Cebula, T.A.; Mammel, M.K.; Feng, P.; et al. Whole-genome draft sequences of 26 enterohemorrhagic Escherichia coli O157:H7 strains. Genome Announc. 2013, 1, e0013412. [Google Scholar] [CrossRef] [PubMed]
  16. Haston, J.C.; Miko, S.; Cope, J.R.; McKeel, H.; Walters, C.; Joseph, L.A.; Griswold, T.; Katz, L.S.; Andujar, A.A.; Tourdot, L.; et al. Cronobacter sakazakii infections in two infants linked to powdered infant formula and breast pump equipment—United States, 2021 and 2022. Mmwr-Morb. Mortal. Wkly. Rep. 2023, 72, 223–226. [Google Scholar] [CrossRef] [PubMed]
  17. Armstrong, G.L.; MacCannell, D.R.; Taylor, J.; Carleton, H.A.; Neuhaus, E.B.; Bradbury, R.S.; Posey, J.E.; Gwinn, M. Pathogen Genomics in Public Health. N. Engl. J. Med. 2019, 381, 2569–2580. [Google Scholar] [CrossRef] [PubMed]
  18. Leeper, M.M.; Tolar, B.M.; Griswold, T.; Vidyaprakash, E.; Hise, K.B.; Williams, G.M.; Im, S.B.; Chen, J.C.; Pouseele, H.; Carleton, H.A. Evaluation of whole and core genome multilocus sequence typing allele schemes for Salmonella enterica outbreak detection in a national surveillance network, PulseNet USA. Front. Microbiol. 2023, 14, 1254777. [Google Scholar] [CrossRef] [PubMed]
  19. Joseph, L.A.; Griswold, T.; Vidyaprakash, E.; Im, S.B.; Williams, G.M.; Pouseele, H.A.; Hise, K.B.; Carleton, H.A. Evaluation of core genome and whole genome multilocus sequence typing schemes for Campylobacter jejuni and Campylobacter coli outbreak detection in the USA. Microb. Genom. 2023, 9, 001012. [Google Scholar] [CrossRef] [PubMed]
  20. Bogaerts, B.; Van den Bossche, A.; Verhaegen, B.; Delbrassinne, L.; Mattheus, W.; Nouws, S.; Godfroid, M.; Hoffman, S.; Roosens, N.H.C.; De Keersmaecker, S.C.J.; et al. Closing the gap: Oxford nanopore technologies R10 sequencing allows comparable results to illumina sequencing for SNP-based outbreak investigation of bacterial pathogens. J. Clin. Microbiol. 2024, 62, e0157623. [Google Scholar] [CrossRef] [PubMed]
  21. Lerminiaux, N.; Fakharuddin, K.; Mulvey, M.R.; Mataseje, L. Do we still need illumina sequencing data? Evaluating oxford nanopore technologies R10.4.1 flow cells and the rapid v14 library prep kit for gram negative bacteria whole genome assemblies. Can. J. Microbiol. 2024, 70, 178–189. [Google Scholar] [CrossRef] [PubMed]
  22. Sanderson, N.D.; Hopkins, K.M.V.; Colpus, M.; Parker, M.; Lipworth, S.; Crook, D.; Stoesser, N. Evaluation of the accuracy of bacterial genome reconstruction with Oxford Nanopore R10.4.1 long-read-only sequencing. Microb. Genom. 2024, 10, 001246. [Google Scholar] [CrossRef] [PubMed]
  23. Ritchie, G.; Chorlton, S.D.; Matic, N.; Bilawka, J.; Gowland, L.; Leung, V.; Stefanovic, A.; Romney, M.G.; Lowe, C.F. WGS of a cluster of MDR Shigella sonnei utilizing Oxford Nanopore R10.4.1 long-read sequencing. J. Antimicrob. Chemother. 2024, 79, 55–60. [Google Scholar] [CrossRef] [PubMed]
  24. Hong, Y.P.; Chen, B.H.; Wang, Y.W.; Teng, R.H.; Wei, H.L.; Chiou, C.S. The usefulness of nanopore sequencing in whole-genome sequencing-based genotyping of Listeria monocytogenes and Salmonella enterica serovar Enteritidis. Microbiol. Spectr. 2024, 12, e0050924. [Google Scholar] [CrossRef] [PubMed]
  25. FDA. FDA’s Guidelines for the Validation of Analytical Methods Using Nucleic Acid Sequenced-Based Technologies. Available online: https://www.fda.gov/food/laboratory-methods-food/foods-program-methods-validation-processes-and-guidelines (accessed on 30 October 2024).
  26. Petersen, C.; Sorensen, T.; Westphal, K.R.; Fechete, L.I.; Sondergaard, T.E.; Sorensen, J.L.; Nielsen, K.L. High molecular weight DNA extraction methods lead to high quality filamentous ascomycete fungal genome assemblies using Oxford Nanopore sequencing. Microb. Genom. 2022, 8, 000816. [Google Scholar] [CrossRef] [PubMed]
  27. Gand, M.; Bloemen, B.; Vanneste, K.; Roosens, N.H.C.; De Keersmaecker, S.C.J. Comparison of 6 DNA extraction methods for isolation of high yield of high molecular weight DNA suitable for shotgun metagenomics Nanopore sequencing to detect bacteria. BMC Genom. 2023, 24, 438. [Google Scholar] [CrossRef] [PubMed]
  28. Greig, D.R.; Jenkins, C.; Gharbia, S.E.; Dallman, T.J. Analysis of a small outbreak of Shiga toxin-producing Escherichia coli O157:H7 using long-read sequencing. Microb. Genom. 2021, 7, 000545. [Google Scholar] [CrossRef] [PubMed]
  29. Vereecke, N.; Van Hoorde, S.; Sperling, D.; Theuns, S.; Devriendt, B.; Cox, E. Virotyping and genetic antimicrobial susceptibility testing of porcine ETEC/STEC strains and associated plasmid types. Front. Microbiol. 2023, 14, 1139312. [Google Scholar] [CrossRef] [PubMed]
  30. Juraschek, K.; Borowiak, M.; Tausch, S.H.; Malorny, B.; Kasbohrer, A.; Otani, S.; Schwarz, S.; Meemken, D.; Deneke, C.; Hammerl, J.A. Outcome of different sequencing and assembly approaches on the detection of plasmids and localization of antimicrobial resistance genes in commensal Escherichia coli. Microorganisms 2021, 9, 598. [Google Scholar] [CrossRef] [PubMed]
  31. Wick, R.R.; Holt, K.E. Benchmarking of long-read assemblers for prokaryote whole genome sequencing. F1000Research 2019, 8, 2138. [Google Scholar] [CrossRef] [PubMed]
  32. Sanderson, N.D.; Kapel, N.; Rodger, G.; Webster, H.; Lipworth, S.; Street, T.L.; Peto, T.; Crook, D.; Stoesser, N. Comparison of R9.4.1/Kit10 and R10/Kit12 Oxford Nanopore flowcells and chemistries in bacterial genome reconstruction. Microb. Genom. 2023, 9, 000910. [Google Scholar] [CrossRef] [PubMed]
  33. Zhao, W.; Zeng, W.; Pang, B.; Luo, M.; Peng, Y.; Xu, J.; Kan, B.; Li, Z.; Lu, X. Oxford nanopore long-read sequencing enables the generation of complete bacterial and plasmid genomes without short-read sequencing. Front. Microbiol. 2023, 14, 1179966. [Google Scholar] [CrossRef] [PubMed]
  34. Sereika, M.; Kirkegaard, R.H.; Karst, S.M.; Michaelsen, T.Y.; Sorensen, E.A.; Wollenberg, R.D.; Albertsen, M. Oxford Nanopore R10.4 long-read sequencing enables the generation of near-finished bacterial genomes from pure cultures and metagenomes without short-read or reference polishing. Nat. Methods 2022, 19, 823–826. [Google Scholar] [CrossRef] [PubMed]
  35. Kolmogorov, M.; Yuan, J.; Lin, Y.; Pevzner, P.A. Assembly of long, error-prone reads using repeat graphs. Nat. Biotechnol. 2019, 37, 540–546. [Google Scholar] [CrossRef]
  36. Bolger, A.M.; Lohse, M.; Usadel, B. Trimmomatic: A flexible trimmer for Illumina sequence data. Bioinformatics 2014, 30, 2114–2120. [Google Scholar] [CrossRef]
  37. Wick, R.R.; Judd, L.M.; Gorrie, C.L.; Holt, K.E. Unicycler: Resolving bacterial genome assemblies from short and long sequencing reads. PLoS Comput. Biol. 2017, 13, e1005595. [Google Scholar] [CrossRef] [PubMed]
  38. Wood, D.E.; Lu, J.; Langmead, B. Improved metagenomic analysis with Kraken 2. Genome Biol. 2019, 20, 257. [Google Scholar] [CrossRef]
  39. Gangiredla, J.; Rand, H.; Benisatto, D.; Payne, J.; Strittmatter, C.; Sanders, J.; Wolfgang, W.J.; Libuit, K.; Herrick, J.B.; Prarat, M.; et al. GalaxyTrakr: A distributed analysis tool for public health whole genome sequence data accessible to non-bioinformaticians. BMC Genom. 2021, 22, 114. [Google Scholar] [CrossRef] [PubMed]
  40. Joensen, K.G.; Tetzschner, A.M.; Iguchi, A.; Aarestrup, F.M.; Scheutz, F. Rapid and easy in silico serotyping of Escherichia coli isolates by use of whole-genome sequencing data. J. Clin. Microbiol 2015, 53, 2410–2426. [Google Scholar] [CrossRef] [PubMed]
  41. Bortolaia, V.; Kaas, R.S.; Ruppe, E.; Roberts, M.C.; Schwarz, S.; Cattoir, V.; Philippon, A.; Allesoe, R.L.; Rebelo, A.R.; Florensa, A.F.; et al. ResFinder 4.0 for predictions of phenotypes from genotypes. J. Antimicrob. Chemother. 2020, 75, 3491–3500. [Google Scholar] [CrossRef] [PubMed]
  42. Nei, M.; Tajima, F.; Tateno, Y. Accuracy of estimated phylogenetic trees from molecular data. II. Gene frequency data. J. Mol. Evol. 1983, 19, 153–170. [Google Scholar] [CrossRef] [PubMed]
Figure 1. NGS validation workflow for pure bacterial isolates according to the Guidelines for the Validation of Analytical Methods Using Nucleic Acid Sequenced-Based Technologies from the FDA (https://www.fda.gov/food/laboratory-methods-food/foods-program-methods-validation-processes-and-guidelines, accessed on 30 October 2024).
Figure 1. NGS validation workflow for pure bacterial isolates according to the Guidelines for the Validation of Analytical Methods Using Nucleic Acid Sequenced-Based Technologies from the FDA (https://www.fda.gov/food/laboratory-methods-food/foods-program-methods-validation-processes-and-guidelines, accessed on 30 October 2024).
Ijms 25 11877 g001
Figure 2. Results of the wgMLST analyses for CFSAN000189. (A) Snapshot of the NJ tree generated from the wgMLST analysis of the ONT assemblies obtained by the different combinations tested against a set of known genomes closely related to that same strain (ST909). (B) Minimum spanning tree (MST) showing the differences between the different CFSAN000189 assemblies obtained by different sequencing technologies. The complete NJ tree can be found in Supplementary Figure S3.
Figure 2. Results of the wgMLST analyses for CFSAN000189. (A) Snapshot of the NJ tree generated from the wgMLST analysis of the ONT assemblies obtained by the different combinations tested against a set of known genomes closely related to that same strain (ST909). (B) Minimum spanning tree (MST) showing the differences between the different CFSAN000189 assemblies obtained by different sequencing technologies. The complete NJ tree can be found in Supplementary Figure S3.
Ijms 25 11877 g002
Figure 3. Results of the wgMLST analyses for CFSAN123154. (A) NJ tree generated from the wgMLST analysis of the ONT assemblies obtained by the different combinations tested against a set of known genomes closely related to that same strain. (B) Minimum spanning tree (MST) showing the differences between the different CFSAN123154 assemblies obtained by different sequencing technologies. The complete NJ tree can be found in Supplementary Figure S4.
Figure 3. Results of the wgMLST analyses for CFSAN123154. (A) NJ tree generated from the wgMLST analysis of the ONT assemblies obtained by the different combinations tested against a set of known genomes closely related to that same strain. (B) Minimum spanning tree (MST) showing the differences between the different CFSAN123154 assemblies obtained by different sequencing technologies. The complete NJ tree can be found in Supplementary Figure S4.
Ijms 25 11877 g003
Figure 4. Results of the wgMLST analyses for CFSAN030807. (A) Snapshot of the NJ tree generated from the wgMLST analysis of the ONT assemblies obtained by the different combinations tested against a set of known genomes (156) closely related to that same strain (ST152). (B) Minimum spanning tree (MST) showing the differences between the different CFSAN030807 assemblies obtained by different sequencing technologies. The complete NJ tree can be found in Supplementary Figure S5.
Figure 4. Results of the wgMLST analyses for CFSAN030807. (A) Snapshot of the NJ tree generated from the wgMLST analysis of the ONT assemblies obtained by the different combinations tested against a set of known genomes (156) closely related to that same strain (ST152). (B) Minimum spanning tree (MST) showing the differences between the different CFSAN030807 assemblies obtained by different sequencing technologies. The complete NJ tree can be found in Supplementary Figure S5.
Ijms 25 11877 g004
Figure 5. Results of the wgMLST analyses for CFSAN076620. (A) NJ tree generated from the wgMLST analysis of the ONT assemblies obtained by the different combinations tested against a set of known genomes (54) closely related to that same strain (ST629). (B) Minimum spanning tree (MST) showing the differences between the different CFSAN076620 assemblies obtained by different sequencing technologies.
Figure 5. Results of the wgMLST analyses for CFSAN076620. (A) NJ tree generated from the wgMLST analysis of the ONT assemblies obtained by the different combinations tested against a set of known genomes (54) closely related to that same strain (ST629). (B) Minimum spanning tree (MST) showing the differences between the different CFSAN076620 assemblies obtained by different sequencing technologies.
Ijms 25 11877 g005
Figure 6. Results of the wgMLST analyses for CFSAN086181. (A) Snapshot of the NJ tree generated from the wgMLST analysis of the ONT assemblies obtained by the different combination tested against a set of known genomes (54) closely related to that same strain (ST629). (B) Minimum spanning tree (MST) showing the differences between the different CFSAN086181 assemblies obtained by different sequencing technologies. The complete NJ tree can be found in Supplementary Figure S6.
Figure 6. Results of the wgMLST analyses for CFSAN086181. (A) Snapshot of the NJ tree generated from the wgMLST analysis of the ONT assemblies obtained by the different combination tested against a set of known genomes (54) closely related to that same strain (ST629). (B) Minimum spanning tree (MST) showing the differences between the different CFSAN086181 assemblies obtained by different sequencing technologies. The complete NJ tree can be found in Supplementary Figure S6.
Ijms 25 11877 g006
Table 1. Verification strains set used in this single laboratory evaluation study.
Table 1. Verification strains set used in this single laboratory evaluation study.
CFSANOrganismSerovarAccession No. a
CFSAN000189Salmonella enterica subsp. entericaBareillyNZ_CP034177.1, NZ_CP034178.1
CFSAN123154Vibrio parahemolyticus Our own data
CFSAN030807Shigella sonnei NZ_CP023645.1, NZ_CP023646.1, NZ_CP023647.1, NZ_CP023648.1, NZ_CP023649.1, NZ_CP023650.1, NZ_CP023651.1, NZ_CP023652.1, NZ_CP023653.1
CFSAN076620Escherichia coliO157:H7Our own data
CFSAN086181Klebsiella pneumoniae NZ_CP045782
a at NCBI.
Table 2. Initial testing of the Q20+ chemistry (RBK114 or LSK114 with R10.4.1 flow cells).
Table 2. Initial testing of the Q20+ chemistry (RBK114 or LSK114 with R10.4.1 flow cells).
RunsFlowcell No.DNA Extraction KitLibrary Prep Kit aBasecalling bOutput Read No.
1FAW43075MaxwellSQK-LSK114SUP 260 bps10.65 Gb1.76 M
2FAW25027MaxwellSQK-RBK114.24SUP 260 bps4.3 Gb705.61k
3FAW42567NEBSQK-LSK114SUP 260 bps8.83 Gb879.33k
4FAW18242NEBSQK-RBK114.24SUP 260 bps3.28 Gb305.31k
Maxwell—Maxwell RSC Cultured Cell DNA kit, NEB—Monarch high molecular weight (HMW). a We used 50 ng per strain per flow cell in the case of the RBK kit and 400 ng for the LSK, as recommended by the manufacturer’s protocols for each kit. b SUP—Super-accurate, MinKNOW software version 22.12.5, basecall_model—[email protected].
Table 3. Assembly statistics showing genome size and coverage for each sample sequenced using any of the four combinations tested.
Table 3. Assembly statistics showing genome size and coverage for each sample sequenced using any of the four combinations tested.
Total Assembly Length (bp) Coverage (X)
SampleRef. Genome Size (bp)FAW25027FAW18242FAW43075FAW42567FAW25027FAW18242FAW43075FAW42567
a CFSAN000189 4,808,5214,806,5924,806,6184,813,5264,806,59892145357305
b CFSAN1231545,021,3035,043,5925,069,0965,053,9595,037,1195841174405
c CFSAN030807 4,848,2804,861,9244,869,4104,866,7044,831,50221063329180
d CFSAN0766205,443,9035,465,8875,442,1565,465,2745,445,8844250277141
e CFSAN0861815,221,6905,221,9055,222,4235,225,4425,221,8677971344184
CFSAN000189 (Salmonella Bareilly), CFSAN123154 (Vibrio parahemolyticus), CFSAN030807 (Shigella sonnei), CFSAN076620 (E. coli O157:H7), and CFSAN086181 (Klebsiella pneumoniae). Genome references at NCBI if available: a CP045753.1 and CP045752.1, b our own data, c GCA_002442535.1, d our own data, and e NZ_CP045782.
Table 4. Assembly statistics showing genome completeness and contig number for each sample sequenced using any of the four combinations tested.
Table 4. Assembly statistics showing genome completeness and contig number for each sample sequenced using any of the four combinations tested.
Genome Completeness Contig Number
SampleFAW25027FAW18242FAW43075FAW42567FAW43075FAW42567FAW25027FAW18242
a CFSAN000189 YYYY2222
b CFSAN123154YYYY5346
c CFSAN030807 YYYY6667
d CFSAN076620YNYN3238
e CFSAN086181YNYY2111
Y means completely closed and N means incomplete chromosome as reported by Flye. Genome references at NCBI if available: a CP045753.1 and CP045752.1, b TBD, c GCA_002442535.1, d TBD, and e NZ_CP045782.1. Expected contig number: Salm (2), Vp (4), Shig (8), E. coli (2), and Kleb (1).
Table 5. Percentage of ORF recovered using any of the testing combinations for each strain per run. Indels and missing genes are also shown.
Table 5. Percentage of ORF recovered using any of the testing combinations for each strain per run. Indels and missing genes are also shown.
% ORF Genes Recovered Intact Using wgMLST Scheme (Genes Missing/Indels)
SamplesFAW43075FAW25027FAW42567FAW18242Total Genes a
CFSAN000189 99.2 (1/30)99.1 (2/37)99.1 (1/36)99.3 (0/30)4296
CFSAN12315499.5 (2/52)99.2 (5/57)99.5 (9/44)98.9 (2/79)4440
CFSAN03080799.8 (2/8)100 (1/0)99.8 (1/6)99.7 (2/11)3669
CFSAN07662099.7 (1/34)99.6 (4/35)99.8 (1/31)99.5 (8/66)4639
CFSAN08618199.6 (2/21)99.8 (0/10)99.7 (1/15)99.7 (4/20)4841
a Total genes in each wgMLST scheme (reference genome specific) used for the analysis.
Table 6. SNP matrix of the second wgMLST analysis for CFSAN000189, showing SNP differences among the different combinations of ONT sequencing and the PacBio and Hybrid assemblies.
Table 6. SNP matrix of the second wgMLST analysis for CFSAN000189, showing SNP differences among the different combinations of ONT sequencing and the PacBio and Hybrid assemblies.
123456789
1CFSAN000189_FAW25027_f0119133131100100100100100
2CFSAN000189_FAW42567_f1190143137109109109109109
3CFSAN000189_FAW43075_f1331430126124124124124124
4CFSAN000189_FAW18242_f1311371260117117117117117
5CFSAN000189 CP04575310010912411700000
6CFSAN000189_M110010912411700000
7CFSAN000189_M210010912411700000
8CFSAN000189_M310010912411700000
9CFSAN000189_NCBI10010912411700000
NCBI—PacBio, CP045753—Hybrid ONT + Illumina, M—MiSeq.
Table 7. SNP matrix of the second wgMLST analysis for CFSAN123154, showing SNP differences among the different combinations of ONT sequencing and the PacBio assembly.
Table 7. SNP matrix of the second wgMLST analysis for CFSAN123154, showing SNP differences among the different combinations of ONT sequencing and the PacBio assembly.
12345678
1CFSAN123154_FAW25027_f0159101102101101138153
2CFSAN123154_FAW18242_f1590130131130130160171
3CFSAN123154_M21011300102112136
4CFSAN123154_M11021311013113137
5CFSAN123154_M31011300102112136
6CFSAN123154_P1011302320112136
7CFSAN123154_FAW43075_f1381601121131121120131
8CFSAN123154_FAW42567_f1531711361371361361310
P—PacBio, M—MiSeq.
Table 8. SNP matrix of the second wgMLST analysis for CFSAN030807, showing SNP differences among the different combinations of ONT sequencing and the PacBio assembly.
Table 8. SNP matrix of the second wgMLST analysis for CFSAN030807, showing SNP differences among the different combinations of ONT sequencing and the PacBio assembly.
12345678
1CFSAN03080702020000
2CFSAN030807_FAW18242_f20242222
3CFSAN030807_FAW25027_f02020000
4CFSAN030807_FAW42567_f24202222
5CFSAN030807_FAW43075_f02020000
6CFSAN030807_M102020000
7CFSAN030807_M302020000
8CFSAN030807_P02020000
P—PacBio, M—MiSeq.
Table 9. SNP matrix of the second wgMLST analysis for CFSAN076620, showing SNP differences among the different combinations of ONT sequencing and the PacBio assembly.
Table 9. SNP matrix of the second wgMLST analysis for CFSAN076620, showing SNP differences among the different combinations of ONT sequencing and the PacBio assembly.
123456
1CFSAN076620_FAW18242_f02836283035
2CFSAN076620_M12809329
3CFSAN076620_FAW42567_f369012119
4CFSAN076620_FAW25027_f283120510
5CFSAN076620_P302115011
6CFSAN076620_FAW43075_f359910110
P—PacBio, M—MiSeq.
Table 10. SNP matrix of the second wgMLST analysis for CFSAN086181, showing SNP differences among the different combinations of ONT sequencing and the PacBio, Hybrid, and Illumina assemblies.
Table 10. SNP matrix of the second wgMLST analysis for CFSAN086181, showing SNP differences among the different combinations of ONT sequencing and the PacBio, Hybrid, and Illumina assemblies.
12345678910
1LEMB18 NZ_CP045782.1_hyb0000119340
2CFSAN086181_M10000119340
3CFSAN086181_M20000119340
4CFSAN086181_M30000119340
545006_SRR904285711110210451
6CFSAN086181_FAW25027_f1111208451
7CFSAN086181_FAW18242_f9999108010119
8CFSAN086181_FAW42567_f33334410053
9CFSAN086181_FAW43075_f44445511504
10CFSAN086181_P0000119340
P—PacBio, M—MiSeq, hyb—Hybrid ONT + Illumina.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Hoffmann, M.; Jang, J.H.; Tallent, S.M.; Gonzalez-Escalona, N. Single Laboratory Evaluation of the Q20+ Nanopore Sequencing Kit for Bacterial Outbreak Investigations. Int. J. Mol. Sci. 2024, 25, 11877. https://doi.org/10.3390/ijms252211877

AMA Style

Hoffmann M, Jang JH, Tallent SM, Gonzalez-Escalona N. Single Laboratory Evaluation of the Q20+ Nanopore Sequencing Kit for Bacterial Outbreak Investigations. International Journal of Molecular Sciences. 2024; 25(22):11877. https://doi.org/10.3390/ijms252211877

Chicago/Turabian Style

Hoffmann, Maria, Jay Hee Jang, Sandra M. Tallent, and Narjol Gonzalez-Escalona. 2024. "Single Laboratory Evaluation of the Q20+ Nanopore Sequencing Kit for Bacterial Outbreak Investigations" International Journal of Molecular Sciences 25, no. 22: 11877. https://doi.org/10.3390/ijms252211877

APA Style

Hoffmann, M., Jang, J. H., Tallent, S. M., & Gonzalez-Escalona, N. (2024). Single Laboratory Evaluation of the Q20+ Nanopore Sequencing Kit for Bacterial Outbreak Investigations. International Journal of Molecular Sciences, 25(22), 11877. https://doi.org/10.3390/ijms252211877

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop