The IR Biotyper is a new automated typing system based on Fourier-transform infrared (FT-IR) spectroscopy that gives results within 4 h. We aimed (i) to use the IR Biotyper to retrospectively analyze an outbreak of extended-spectrum beta-lactamase-producing Klebsiella pneumoniae (ESBL-KP) in a neonatal intensive care unit and to compare results to BOX-PCR and whole-genome sequencing (WGS) results as the gold standard and (ii) to assess how the cutoff values used to define clusters affect the discriminatory power of the IR Biotyper.
KEYWORDS: FT-IR spectroscopy, IR Biotyper, K. pneumoniae, NICU, outbreak, WGS, BOX-PCR, ESBL
ABSTRACT
The IR Biotyper is a new automated typing system based on Fourier-transform infrared (FT-IR) spectroscopy that gives results within 4 h. We aimed (i) to use the IR Biotyper to retrospectively analyze an outbreak of extended-spectrum beta-lactamase-producing Klebsiella pneumoniae (ESBL-KP) in a neonatal intensive care unit and to compare results to BOX-PCR and whole-genome sequencing (WGS) results as the gold standard and (ii) to assess how the cutoff values used to define clusters affect the discriminatory power of the IR Biotyper. The sample consisted of 18 isolates from 14 patients. Specimens were analyzed in the IR Biotyper using the default analysis settings, and spectra were analyzed using OPUS 7.5 software. The software contains a feature that automatically proposes a cutoff value to define clusters; the cutoff value defines up to which distance the spectra are considered to be in the same cluster. Based on FT-IR, the outbreak represented 1 dominant clone, 1 secondary clone, and several unrelated clones. FT-IR results, using the cutoff value generated by the accompanying software after 4 replicates, were concordant with WGS for all but 1 isolate. BOX-PCR was underdiscriminatory compared to the other two methods. Using the cutoff value generated after 12 replicates, the results of FT-IR and WGS were completely concordant. The IR Biotyper can achieve the same typeability and discriminatory power as genome-based methods. However, to attain this high performance requires either previous, strain-dependent knowledge about the optimal technical parameters to be used or validation by a second method.
INTRODUCTION
Third-generation cephalosporin-resistant Enterobacteriaceae are listed in the “critical” category in the WHO’s list of priority pathogens requiring new antibiotics (1). Extended spectrum beta-lactamase- (ESBL)-producing Klebsiella pneumoniae (ESBL-KP) is an important cause of nosocomial outbreaks in neonatal intensive care units (NICUs) (2). ESBL-KP has a wide range of clinical manifestations in neonates, including urinary tract infections, pneumonia, sepsis, and meningitis (2, 3).
Typing of bacterial strains can aid in outbreak investigation by revealing clonality and possible routes of transmission and by linking clinical isolates to environmental reservoirs (4). Typing methods include those that rely on genomic fingerprinting methods using DNA amplification, such as REP-PCR, BOX-PCR, and ERIC-PCR (5, 6), DNA-based methods that do not require amplification, such as pulsed-field gel electrophoresis (PFGE) (7), whole-genome sequencing (WGS) (8, 9), and methods that are not DNA based, such as matrix-assisted laser desorption–ionization time-of-flight mass spectrometry (MALDI-TOF MS) (10, 11). Typeability, ease of use, ease of interpretation, rapidity, reproducibility, discriminatory power, cost, and available expertise are important considerations when choosing a typing method (4, 12). For example, despite the excellent reproducibility and discriminatory power of WGS, it is still far from being routinely used because of its high cost, turnaround time to reporting (a few days at best or weeks if outsourced), and the expertise required to analyze the results (13).
Recently, a new automated typing system based on Fourier-transform infrared (FT-IR) spectroscopy, the IR Biotyper (Bruker GmbH, Bremen, Germany), became commercially available (14). FT-IR spectroscopy is a spectrum-based technique, first introduced in the 1950s, that quantifies the absorption of infrared light by molecules present in the bacterial cell. The IR spectrum generated provides a specific fingerprint that reflects the cell composition of nucleic acids, proteins, lipids, and carbohydrates. This results in the generation of a specific IR spectrum, which reflects the overall chemical composition of the specimen. Thus, each microorganism has a highly specific infrared absorption signature correlated with genetic information allowing the identification of microorganisms on the subspecies level (15, 16).
The aims of this study were (i) to use the IR Biotyper to retrospectively analyze a NICU outbreak of ESBL-KP and to compare the results to BOX-PCR and WGS (the gold standard) and (ii) to assess how the cutoff values used to define clusters affect the discriminatory power of the IR Biotyper.
MATERIALS AND METHODS
Bacterial strains.
The sample included 18 ESBL-KP isolates from 14 patients (neonates and mothers) collected during an outbreak in an Israeli NICU in July to August 2017. The sample sources were blood (n = 2), cerebrospinal fluid (CSF) (n = 1), birth canal (n = 1), wound (n = 1), and screening rectal swabs (n = 13). One ESBL-KP isolate not related to the outbreak was included as a control. All isolates were identified to the species level using Vitek MS (bioMérieux SA, Marcy l’Etoile, France), and antibiotic susceptibility was determined using Vitek 2 (bioMérieux SA, Marcy l’Etoile, France). Confirmatory testing for ESBL production was carried out with the double-disc diffusion test in accordance with CLSI guidelines (17). The isolates were stored at –80°C and subcultured before testing.
FT-IR spectroscopy methods.
Sample preparation. Isolates were grown at 35 ± 2°C on blood agar plates (tryptic soy agar supplemented with 5% sheep blood; Hylabs, Rehovot, Israel) for 24 h. Samples were prepared according to the IR Biotyper manufacturer’s instructions. An inoculation loopful (approximately 10 μl) of bacteria was suspended in 50 μl of 70% ethanol solution (Bio-Lab, Jerusalem, Israel). Homogenization was obtained by vortexing the sample with metal beads (Bruker, Bremen, Germany) in 1.5-ml microcentrifuge tubes. An additional volume (50 μl) of water (molecular grade; Bio-Lab, Jerusalem, Israel) was added to obtain a solution with a final volume of 100 μl. An aliquot (15 μl) of the suspension was placed on a sample plate device (Bruker, Bremen, Germany) and dried at 37°C for 30 min. The pellets were visualized to check for the absence of air bubbles before analysis.
Sample analysis and spectrum acquisition. The specimens were analyzed in quadruplicate in three independent experiments using the IR Biotyper with the default analysis settings. Spectra were analyzed using OPUS 7.5 software (Bruker GmbH, Bremen, Germany). Quality control was performed according to the manufacturer’s recommendations and based on several criteria, such as absorption (0.4 arbitrary units [AU] < D value < 2 AU), noise (<150 × 10−6 AU), presence of water (<300 × 10−6 AU), and fringes (<100 × 10−6 AU). Samples that failed to meet the quality control criteria were excluded from the analysis.
Spectrum analysis. Hierarchical cluster analysis (HCA) was generated with the OPUS 7.5 software using the Pearson correlation coefficient option. The software contains a feature that automatically proposes a cutoff value to define clusters; the cutoff value defines up to which distance the spectra are considered to be in the same cluster. The cutoff value generated by the software changes depending on the number of technical replicates. To date, the company has not issued validated cutoff values, but the company does provide a table of recommended cutoff values that has been compiled based on input from users. The cutoff values differ by organism.
Whole-genome sequencing methods. Genomic DNA was sequenced at the Sequencing Core-UIC RRC (University of Illinois at Chicago). DNA libraries were prepared using a Nextera XT kit (Illumina, Inc., San Diego, CA) according to the manufacturer’s instructions and sequenced on the NextSeq 500 instrument (Illumina, Inc.).
Bioinformatics analysis. All bioinformatics analyses were performed at our laboratory at the National Institute for Antibiotic Resistance and Infection Control. Adapters and low-quality sequences were removed after screening of raw sequences. Sequence data were analyzed using a publicly available program from the Center for Genomic Epidemiology to determine multilocus sequence typing (MLST) types (https://cge.cbs.dtu.dk/services/MLST/). Genes used for core genome alignment were derived from pan-genome analysis using Roary software 3.12.0. The resulting sequences were aligned using default settings of the MAFFT multiple sequence alignment program 7.407, based on nucleotide polymorphism. A maximum likelihood phylogenetic tree was inferred using RAxML 8.2.12 with a gamma model of rate heterogeneity with 100 bootstrap replicates (18).
BOX-PCR method. Isolates were grown at 37°C on blood agar plates (Hylabs, Rehovot, Israel) for 24 h prior to analysis. We used the BOX-PCR method described by Louws et al. (19). PCR products were run using capillary gel electrophoresis (QIAxcel; Qiagen, Inc., CA, USA) and were visualized and analyzed using QIAxcel ScreenGel 1.2.0. The genomic profile data were analyzed using R statistical software 3.4.2 (https://www.r-project.org/). A Pearson’s product-moment correlation coefficient was used to calculate the samples’ distance matrix using the R package factoextra 1.0.5 (http://www.rdocumentation.org/packages/factoextra). Dendrograms were constructed using DendroUPGMA (20).
Discriminatory power and concordance. To compare the discriminatory ability of each of the three typing systems, the discrimination index (D) was calculated (21). D describes the ability of a test to distinguish between two unrelated strains. A value of 1 indicates that all strains are different, and 0 indicates that all strains are identical. The concordance of different typing methods was calculated using the adjusted Wallace coefficient (http://www.comparingpartitions.info/index.php?link=Home). A value of 1 indicates perfect agreement between methods, and 0 indicates no agreement.
Data availability.
Draft genome sequence data for the strains from this study were deposited in the National Center for Biotechnology (NCBI) with accession numbers SAMN11270017 to SAMN11270024.
RESULTS
Comparison of the three typing methods.
The FT-IR spectroscopy analysis is presented in Fig. 1. FT-IR spectroscopy was initially performed using the cutoff value of 0.037, which was automatically generated by the software’s algorithm based on analysis of 4 technical replicates. Nine isolates (I1 and I3 to I10) from six patients (patients P1 to P6) were classified as belonging to the predominant outbreak clone (clone I). The second isolate (I2), from patient P1’s rectal swab, was classified as a singleton (clone II). Three additional isolates (I11 to I13) from patients P7 to P9 formed a secondary cluster (clone III). The remaining five isolates were unrelated (clones IV to VIII). To confirm our analysis, a control ESBL-KP strain (I19) was included, and the IR Biotyper classified it as a distinct singleton (IX). The index of discrimination was 0.719 (95% confidence interval [CI]. 0.516 to 0.923).
FIG 1.
IR biotyping of ESBL-producing K. pneumoniae isolates. Dendrogram obtained by IR biotyping analysis of ESBL-producing K. pneumoniae specimens (n = 18) isolated during an outbreak that occurred in a NICU. Analysis was based on four replicates for each isolate. The blue line represents the automatic cutoff value (0.037). For ease of reading, clusters composed by two isolates or more are shaded orange; singletons are shaded green.
A comparison of the results of FT-IR spectroscopy, BOX-PCR, and WGS is described in Table 1. BOX-PCR identified two clusters. Cluster A was composed of 10 isolates (I1 to I10) from six patients (P1 to P6); it was identical to clone I obtained using the IR Biotyper. Unlike the IR Biotyper, BOX-PCR did not differentiate between the 4 isolates from patient P1 (I1 to I4) and clustered them all under the major cluster A. Cluster B was composed of four isolates (I11 to I14) from four patients (P7 to P10); these corresponded to the two strains in IR biotyping clone III and the two singleton clones IV and V. The remaining four isolates (I15 to I18) were classified as unrelated, as they were in the IR biotyping classification. The index of discrimination for BOX-PCR was 0.702 (95% CI, 0.510 to 0.893). The adjusted Wallace coefficient indicating the concordance between the BOX-PCR and FT-IR spectroscopy methods was 0.650 (95% CI, 0.296 to 1.000).
TABLE 1.
Comparison of IR biotyping to WGS and BOX-PCR fingerprintinga
Patient no. | Isolate no. | Source | Isolation date (day.mo.yr) | Classification using: |
|||
---|---|---|---|---|---|---|---|
IR Biotyper (clone no.)d | cgMLST (ST) | Core genome alignment (clone no.) | BOX-PCR (cluster) | ||||
1 | 1 | Blood | 18.07.17 | I | ST985 | 1 | A |
2 | Rectal | 21.07.17 | II | ST985 | 1 | A | |
3 | Wound | 18.07.17 | I | ST985 | 1 | A | |
4 | CSF | 18.07.17 | I | ST985 | 1 | A | |
2 | 5 | Rectal | 21.07.17 | I | ST985 | 1 | A |
6 | Blood | 21.07.17 | I | ST985 | 1 | A | |
3 | 7 | Rectal | 26.07.17 | I | ST985 | 1 | A |
4 | 8 | Blood | 17.07.17 | I | ST985 | 1 | A |
5 | 9 | Rectal | 21.07.17 | I | ST985 | 1 | A |
6 | 10 | Rectal | 26.07.17 | I | ST985 | 1 | A |
7 | 11 | Rectal | 21.07.17 | III | ST464 | 2 | B |
8 | 12 | Rectal | 21.07.17 | III | ST464 | 2 | B |
9 | 13 | Rectal | 30.07.17 | III | ST464 | 2 | B |
10 | 14 | Rectal | 21.07.17 | IV | ST13 | 3 | B |
11 | 15 | Rectal | 27.07.17 | V | ST16 | 4 | C |
12 | 16c | Rectal | 30.07.17 | VI | ST480 | 5 | D |
13 | 17 | Rectal | 31.07.17 | VII | ST46 | 6 | E |
14 | 18 | Rectal | 27.07.17 | VIII | ST36 | 7 | F |
Control | 19 | Environmental | 16.08.17 | IX | ST2286 | 8 | G |
Statistics | |||||||
Discrimination indexb (95% CI) | 0.784 (0.595–0.972) | 0.719 (0.516–0.923) | 0.719 (0.516–0.923) | 0.702 (0.510–0.893) | |||
Adjusted Wallace coefficiente (95% CI) | 0.708 (0.335–1.000) | 0.708 (0.335–1.000) | 0.650 (0.296–1.000) |
IR biotyping analysis of ESBL-producing K. pneumoniae specimens (n = 18) isolated during an outbreak that occurred in a NICU was compared to cgMLST, core genome alignment, and BOX-PCR.
Based on Simpson’s index of diversity.
Identified as K. quasipneumoniae using cgMLST and core genome alignment.
Classification based on the automatic cutoff value of 0.037.
Adjusted Wallace coefficients compared to IR Biotyper. Discrepancy between core genome alignment and FT-IR biotyping is shaded.
The results of WGS were concordant with IR biotyping for all but one isolate, I2. Core genome alignment and core genome MLST (cgMLST) analyses confirmed the presence of two major clusters, sequence type 985 (ST985) (I1 to I10) and ST464 (I11 to I13). Five unrelated singletons belonging to sequence types ST13 (I14), ST16 (I15), ST480 (I16), ST46 (I17), and ST36 (I18) corresponded to the IR biotyping unrelated clones IV to VIII. In addition, cgMLST identified isolate I16 (ST480) as Klebsiella quasipneumoniae rather than K. pneumoniae. The index of discrimination for both core genome alignment and cgMLST was 0.719 (95% CI, 0.516 to 0.923). The adjusted Wallace coefficient indicating the concordance between core genome alignment/cgMLST and FT-IR spectroscopy was 0.708 (95% CI, 0.335 to 1.000).
Comparison of IR Biotyper results using different cutoff values.
As noted, the initial analysis using the IR Biotyper was performed using the cutoff value generated by the software based on an analysis of four technical replicates, 0.037. In contrast, unofficial guidelines issued by the manufacturer of the IR Biotyper recommend a cutoff of 0.20 to 0.25 for K. pneumoniae typing. When the analysis was repeated using a cutoff value in the recommended range, isolates I1 to I10 and I15 to I18 were grouped together in a single clone (Table 2). The index of discrimination decreased from 0.719 using the automatically generated cutoff value to 0.624 using the recommended cutoff value.
TABLE 2.
Influence of the cutoff value on sample classification using IR biotypinga
Isolate no. | cgMLST (ST) | IR Biotyper clone no. for cutoff value |
||
---|---|---|---|---|
0.2b | 0.037c | 0.046d | ||
1 | ST985 | I | I | I |
2 | ST985 | I | II | I |
3 | ST985 | I | I | I |
4 | ST985 | I | I | I |
5 | ST985 | I | I | I |
6 | ST985 | I | I | I |
7 | ST985 | I | I | I |
8 | ST985 | I | I | I |
9 | ST985 | I | I | I |
10 | ST985 | I | I | I |
11 | ST464 | II | III | II |
12 | ST464 | II | III | II |
13 | ST464 | II | III | II |
14 | ST13 | III | V | III |
15 | ST16 | I | VI | IV |
16 | ST480 | I | VII | V |
17 | ST46 | I | VIII | VI |
18 | ST36 | I | IX | VII |
19 | ST2286 | IV | X | VIII |
Discrepancies between core genome alignment and FT-IR biotyping are shaded.
Cutoff value suggested by the manufacturer for K. pneumoniae typing.
Automatic cutoff value generated by the manufacturer’s software based on analysis of 4 technical replicates.
Automatic cutoff value generated by the manufacturer’s software based on analysis of 12 technical replicates.
When the initial automatic cutoff value of 0.037 was used, one isolate (I2) was misclassified compared to the gold standard of WGS. We examined what cutoff value would be needed in order to achieve perfect concordance between the IR Biotyper and WGS. Using 5 to 11 replicates, the concordance did not change. Based on 12 technical replicates, the IR Biotyper software generated a cutoff value of 0.046. Using this cutoff, the results of FT-IR spectroscopy and WGS were identical (Table 2). According to the OPUS 7.5 software, this perfect agreement would be maintained if the cutoff fell in the range of 0.049 to 0.045.
Comparison of practical aspects of the typing methods.
Table 3 compares the practical aspects of performing BOX-PCR fingerprinting, FT-IR biotyping, and WGS. Compared to DNA and/or genome-based approaches, FT-IR is faster, especially in the postanalytical phase. In contrast to BOX-PCR, the HCA does not require processing of graphic information, which is subject to interpretation. In contrast to molecular methods, there was no risk of cross-contamination during the preparation of IR Biotyper samples. A laboratory technician with a basic level of training can perform sample preparation, whereas more qualified personnel are required for molecular or genome sequencing-based methods. The price per sample analyzed using the IR Biotyper was approximately 15 euros, far less expensive than WGS.
TABLE 3.
Overview of practical aspects of BOX-PCR fingerprinting, FT-IR biotyping, and WGS
Method | Time to result (h)a
|
Cost/isolate (euros) | Expertiseb | |||
---|---|---|---|---|---|---|
Preanalytic | Analytic | Postanalytic | Total | |||
BOX-PCR | 2 | 6 | 2 | 10 | 5 | Moderate |
IR Biotyper | 0.5 | 2.5 | 1 | 4 | 15 | Low |
WGSc | 2 | 48–72 | 24 | 74–98 (4 days) | 100–150 | High |
Based on analysis of 20 isolates.
Technical expertise level required from technician.
In-house analysis, including bioinformatics analysis.
DISCUSSION
We retrospectively compared FT-IR biotyping to BOX-PCR and WGS for analyzing isolates from an ESBL-KP outbreak in a NICU. If FT-IR biotyping can provide results with accuracy similar to WGS far more quickly, it could be a valuable tool for real-time outbreak management.
According to the WGS gold standard, the interpretation of the outbreak presented here would be an outbreak caused by 1 major clone (ST985) involving six patients (P1 to P6) and a second, less common clone (ST464, P7 to P9). Would FT-IR results using different cutoff values lead to the same interpretation? Using the cutoff automatically generated after 4 replicates (0.037), the interpretation would be that this was a monoclonal outbreak involving the same 6 patients identified by WGS. One of the four isolates (I2) from P1 was classified as a distinct singleton. This “oversensitivity” does not necessarily represent misclassification; isolates deemed to be genotypically identical can have phenotypic differences detected by FT-IR. For this patient, oversensitivity would not affect outbreak management because he would still be considered part of the main outbreak based on his 3 other isolates.
With the recommended cutoff (0.2), 10 patients (P1 to P6 and P15 to P18) would have been identified by FT-IR as carrying the major clone. This misclassification (undersensitivity) would have complicated the outbreak investigation, as infection control staff would have searched unnecessarily for contacts or a common source of transmission connecting P15 to P18 to the outbreak. Applying this cutoff, the 3 patients with ST464 were grouped correctly.
We reached complete concordance with WGS using the cutoff automatically generated by increasing the number of technical replicates to 12. How many replicates should be performed in order to generate the best cutoff? It is not possible to set definitive rules; there will be differences based on bacterial species, sample size and composition (e.g., outbreak versus unrelated strains), and the epidemiological data available to assist with interpretation of results. In general, performing more replicates will increase the accuracy (i.e., avoid under- and oversensitivity) of the FT-IR analysis. We recommend that any laboratory planning to use FT-IR biotyping for a given organism should initially validate the technical and analytical setup and results using a molecular or genomic approach.
In the early 2010s, most studies using FT-IR spectroscopy focused on foodborne pathogens (22–24). Later, the use expanded to other pathogens with clinical or epidemiological importance (25). We found only one article describing the use of Bruker GmbH’s IR Biotyper for outbreak investigation (26); Martak et al. retrospectively analyzed isolates from outbreaks caused by four Gram-negative pathogens and found that the IR Biotyper results were highly congruent with results obtained with multilocus sequence typing (MLST) and pulsed-field gel electrophoresis (PFGE). Two studies have reported high discriminatory power of the IR Biotyper compared to WGS (27, 28).
Our study has several limitations. We examined the effect of the number of technical replicates on the discriminatory power of FT-IR; we did not test other factors that influence FT-IR performance, such as environmental temperature and humidity, medium, and incubation time. This was a retrospective analysis of isolates that were already known, by epidemiological investigation, to be part of an outbreak. We did not demonstrate whether FT-IR could be integrated into routine surveillance to warn, in real time, that an outbreak is occurring.
In summary, we found that the FT-IR Biotyper can achieve the same typeability and discriminatory power as genome-based methods. However, to attain this high performance requires either previous, strain-dependent knowledge about the optimal technical parameters to be used or validation by a second method. Because of its technical simplicity, short turnaround time, and low cost, FT-IR biotyping has the potential to become an essential tool for outbreak investigation.
ACKNOWLEDGMENTS
Y.C. received research funding from MSD, AstraZeneca, Allecra Therapeutics, DaVoltera, bioMérieux SA, Nariva, Achoagen, Roche, Pfizer, Shionogi, VenatoRx, and Qpex Pharmaceuticals. All other authors report no potential conflicts of interest with respect to the research, authorship, and publication of this article.
REFERENCES
- 1.Tacconelli E, Carrara E, Savoldi A, Harbarth S, Mendelson M, Monnet DL, Pulcini C, Kahlmeter G, Kluytmans J, Carmeli Y, Ouellette M, Outterson K, Patel J, Cavaleri M, Cox EM, Houchens CR, Grayson ML, Hansen P, Singh N, Theuretzbacher U, Magrini N, WHO Pathogens Priority List Working Group . 2018. Discovery, research, and development of new antibiotics: the WHO priority list of antibiotic-resistant bacteria and tuberculosis. Lancet Infect Dis 18:318–327. doi: 10.1016/S1473-3099(17)30753-3. [DOI] [PubMed] [Google Scholar]
- 2.Stapleton PJM, Murphy M, Mccallion N, Brennan M, Cunney R, Drew RJ. 2016. Outbreaks of extended spectrum beta-lactamase-producing Enterobacteriaceae in neonatal intensive care units: a systematic review. Arch Dis Child Fetal Neonatal Ed 101:72–78. doi: 10.1136/archdischild-2015-308707. [DOI] [PubMed] [Google Scholar]
- 3.Hendrik TC, Voor AF, Vos MC. 2015. Clinical and molecular epidemiology of extended-spectrum beta-lactamase-producing Klebsiella spp.: a systematic review and meta-analyses. PLoS One 10:e0140754. doi: 10.1371/journal.pone.0140754. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Foxman B, Zhang L, Koopman JS, Manning SD, Marrs CF. 2005. Choosing an appropriate bacterial typing technique for epidemiologic studies. Epidemiol Perspect Innov 2:10. doi: 10.1186/1742-5573-2-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hashemi A, Baghbani-Arani F. 2015. The effective differentiation of Salmonella isolates using four PCR-based typing methods. J Appl Microbiol 118:1530–1540. doi: 10.1111/jam.12805. [DOI] [PubMed] [Google Scholar]
- 6.Lin CW, Chiou CS, Chang YC, Yang TC. 2008. Comparison of pulsed-field gel electrophoresis and three rep-PCR methods for evaluating the genetic relatedness of Stenotrophomonas maltophilia isolates. Lett Appl Microbiol 47:393–398. doi: 10.1111/j.1472-765X.2008.02443.x. [DOI] [PubMed] [Google Scholar]
- 7.Lopez-Canovas L, Martinez Benitez MB, Herrera Isidron JA, Flores Soto E. 2019. Pulsed field gel electrophoresis: past, present, and future. Anal Biochem 573:17–29. doi: 10.1016/j.ab.2019.02.020. [DOI] [PubMed] [Google Scholar]
- 8.Salipante SJ, Sengupta DJ, Cummings LA, Land TA, Hoogestraat DR, Cookson T. 2015. Application of whole-genome sequencing for bacterial strain typing in molecular epidemiology. J Clin Microbiol 53:1072–1079. doi: 10.1128/JCM.03385-14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Schürch AC, Arredondo-Alonso S, Willems RJL, Goering RV. 2018. Whole genome sequencing options for bacterial strain typing and epidemiologic analysis based on single nucleotide polymorphism versus gene-by-gene-based approaches. Clin Microbiol Infect 24:350–354. doi: 10.1016/j.cmi.2017.12.016. [DOI] [PubMed] [Google Scholar]
- 10.Sauget M, Valot B, Bertrand X, Hocquet D. 2017. Can MALDI-TOF mass spectrometry reasonably type bacteria? Trends Microbiol 25:447–455. doi: 10.1016/j.tim.2016.12.006. [DOI] [PubMed] [Google Scholar]
- 11.Rodríguez-Sánchez B, Cercenado E, Coste AT, Greub G. 2019. Review of the impact of MALDI-TOF MS in public health and hospital hygiene, 2018. Euro Surveill 24:1800193 https://www.eurosurveillance.org/content/10.2807/1560-7917.ES.2019.24.4.1800193. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.van Belkum A, Tassios PT, Dijkshoorn L, Haeggman S, Cookson B, Fry NK, Fussing V, Green J, Feil E, Gerner-Smidt P, Brisse S, Struelens M. 2007. Guidelines for the validation and application of typing methods for use in bacterial epidemiology. Clin Microbiol Infect 13:1–46. doi: 10.1111/j.1469-0691.2007.01786.x. [DOI] [PubMed] [Google Scholar]
- 13.Mintzer V, Moran-Gilad J, Simon-Tuval T. 2019. Operational models and criteria for incorporating microbial whole genome sequencing in hospital microbiology: a systematic literature review. Clin Microbiol Infect 25:1086–1095. doi: 10.1016/j.cmi.2019.04.019. [DOI] [PubMed] [Google Scholar]
- 14.Bruker GmbH Daltonics Division. 2018. Bruker GmbH announces improved solutions for microbial strain typing, hospital hygiene and infection control, and Candida auris testing at ASM 2018. https://www.bruker.com/news/asm/bruker-asm-press-release-2018.html.
- 15.Quintelas C, Ferreira EC, Lopes JA, Sousa C. 2018. An overview of the evolution of infrared spectroscopy applied to bacterial typing. Biotechnol J 13. doi: 10.1002/biot.201700449. [DOI] [PubMed] [Google Scholar]
- 16.Wenning M, Scherer S. 2013. Identification of microorganisms by FTIR spectroscopy: perspectives and limitations of the method. Appl Microbiol Biotechnol 97:7111–7120. doi: 10.1007/s00253-013-5087-3. [DOI] [PubMed] [Google Scholar]
- 17.CLSI. 2019. Performance standards for antimicrobial susceptibility testing, 29th ed CLSI supplement M100 Clinical and Laboratory Standards Institute, Wayne, PA. [Google Scholar]
- 18.Price MN, Dehal PS, Arkin AP. 2010. FastTree 2: approximately maximum-likelihood trees for large alignments. PLoS One 5:e9490. doi: 10.1371/journal.pone.0009490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Louws FJ, Fulbright DW, Stephens CT, de Bruijn FJ. 1994. Specific genomic fingerprints of phytopathogenic Xanthomonas and Pseudomonas pathovars and strains generated with repetitive sequences and PCR. Appl Environ Microbiol 60:2286–2295. doi: 10.1128/AEM.60.7.2286-2295.1994. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Garcia-Vallvé S, Puigbo P. 2002. DendroUPGMA: a dendrogram construction utility. Biochemistry and Biotechnology Department. Rovira i Virgili University, Tarragona, Spain: http://genomes.urv.cat/UPGMA/. [Google Scholar]
- 21.Hunter PR, Gaston MA. 1988. Numerical index of the discriminatory ability of typing systems: an application of Simpson’s index of diversity. J Clin Microbiol 26:2465–2466. doi: 10.1128/JCM.26.11.2465-2466.1988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Fetsch A, Contzen M, Hartelt K, Kleiser A, Maassen S, Rau J, Kraushaar B, Layer F, Strommenger B. 2014. Staphylococcus aureus food-poisoning outbreak associated with the consumption of ice-cream. Int J Food Microbiol 187:1–6. doi: 10.1016/j.ijfoodmicro.2014.06.017. [DOI] [PubMed] [Google Scholar]
- 23.Preisner O, Guiomar R, Machado J, Menezes JC, Lopes JA. 2010. Application of Fourier transform infrared spectroscopy and chemometrics for differentiation of Salmonella enterica serovar Enteritidis phage types. Appl Environ Microbiol 76:3538–3544. doi: 10.1128/AEM.01589-09. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Nyarko EB, Puzey KA, Donnelly CW. 2014. Rapid differentiation of Listeria monocytogenes epidemic clones III and IV and their intact compared with heat-killed populations using Fourier transform infrared spectroscopy and chemometrics. J Food Sci 79:M1189–1196. doi: 10.1111/1750-3841.12475. [DOI] [PubMed] [Google Scholar]
- 25.Novais A, Freitas AR, Rodrigues C, Peixe L. 2019. Fourier transform infrared spectroscopy: unlocking fundamentals and prospects for bacterial strain typing. Eur J Clin Microbiol Infect Dis 38:427–448. doi: 10.1007/s10096-018-3431-3. [DOI] [PubMed] [Google Scholar]
- 26.Martak D, Valot B, Sauget M, Cholley P, Thouverez M, Bertrand X, Hocquet D. 2019. Fourier-transform infrared spectroscopy can quickly type Gram-negative bacilli responsible for hospital outbreaks. Front Microbiol 10:1440. doi: 10.3389/fmicb.2019.01440. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Dinkelacker AG, Vogt S, Oberhettinger P, Mauder N, Rau J, Kostrzewa M, Rossen JWA, Autenrieth IB, Peter S, Liese J. 2018. Typing and species identification of clinical Klebsiella isolates by Fourier transform infrared spectroscopy and matrix-assisted laser desorption ionization–time of flight mass spectrometry. J Clin Microbiol 56:e00843-18. doi: 10.1128/JCM.00843-18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Vogt S, Löffler K, Dinkelacker AG, Bader B, Autenrieth IB, Peter S, Liese J. 2019. Fourier-transform infrared (FTIR) spectroscopy for typing of clinical Enterobacter cloacae complex isolates Front Microbiol 10:1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Draft genome sequence data for the strains from this study were deposited in the National Center for Biotechnology (NCBI) with accession numbers SAMN11270017 to SAMN11270024.