Validation of a simplex PCR assay enabling reliable identification of clinically relevant Candida species

Background Fungal bloodstream infections (BSI) may be serious and are associated with drastic rise in mortality and health care costs. Candida spp. are the predominant etiological agent of fungal sepsis. The prompt and species-level identification of Candida may influence patient outcome and survival. The aim of this study was to develop and evaluate the CanTub-simplex PCR assay coupled with Tm calling and subsequent high resolution melting (HRM) analysis to barcode seven clinically relevant Candida species. Methods Efficiency, coefficient of correlation and the limit of reliable detection were estimated on purified Candida EDTA-whole blood (WB) reference panels seeded with Candida albicans, Candida glabrata, Candida parapsilosis, Candida tropicalis, Candida krusei, Candida guilliermondii, Candida dubliniensis cells in a 6-log range. Discriminatory power was measured on EDTA-WB clinical panels on three different PCR platforms; LightCycler®96, LightCycler® Nano, LightCycler® 2.0. Inter- and intra assay consistencies were also calculated. Results The limit of reliable detection proved to be 0.2–2 genomic equivalent and the method was reliable on broad concentration ranges (106–10 CFU) providing distinctive melting peaks and characteristic HRM curves. The diagnostic accuracy of the discrimination proved to be the best on Roche LightCycler®2.0 platform. Repeatability was tested and proved to be % C.V.: 0.14 ± 0.06 on reference- and % C.V.: 0.14 ± 0.02 on clinical-plates accounting for a very high accuracy. Reproducibility was % C.V.: 0.11 between reference- and % C.V.: 0.12between clinical-panels which is highly acceptable. Conclusion Our assay demonstrates recent advances on Tm calling and HRM analysis for the molecular identification of relevant Candida species. This unique, simplex PCR assay may be capable to outperform conventional phenotypic methods by reducing time and providing accurate and reliable results directly from blood (2 h) or from whole blood culture bottles (12–24 h). Electronic supplementary material The online version of this article (10.1186/s12879-018-3283-6) contains supplementary material, which is available to authorized users.


Background
Fungal bloodstream infections may be serious and are associated with drastic rise in mortality and health care costs [1,2]. Currently, with the gradually growing number of immunocompromised patients, we face an appreciable shift in the spectrum of fungal infections [3,4]. The genus Candida corresponds to the most important cause of opportunistic mycoses in the world [5,6]. Despite the antifungal therapy nosocomial Candida infections constitute a public health problem contributing to prolonged hospitalization time, generating enormous excess of costs for patient treatment, and a high mortality (25-60%) especially when complicated with septic shock [7,8].
Timely diagnosis of sepsis due to candidemia is essential for effective therapy, while delays of more than 12 h in the administration of antifungal drugs may substantially increase mortality [19][20][21][22][23].
The rapid and correct identification of Candida species can narrow therapy options by preventing treatment with potentially toxic antifungal agents, thus reducing costs of hospitalization and improving negative patient outcomes [24][25][26]. Because of the phenotypic similarities of Candida species, the turn-around time of traditional, culture based identification methods may take 2 to 8 days, delay adequate diagnosis and appropriate antifungal treatment. In comparison the surrogate-marker based molecular assays requiring an average of 3-4 h [27]. Nevertheless in some cases the morphological identification of clinically relevant Candida species is hampered by several difficulties, as is the case with the germ tube positive C. albicans and the potentially fluconazole-resistant, cryptic C. dubliniensis providing light green vs. dark green colonies on CHROM agar presenting only subtle differences in colony color [28,29].
With regard to surrogate-marker based molecular methods, the antimannan antibody (Platelia GM-enzyme immunoassay, Bio-Rad) and 1,3-ß-D-glucan (BDG) antigen based (Fungitell assay, Associates of Cape Cod) immunological assays have good performances and may be useful in diagnosing of invasive fungal infections [30]. Nonetheless, as the panfungal BDG is a major cell wall component of Candida and other fungi, except Cryptococci and Zygomycetes, it not suitable for the identification even to the genus level [12].
Along with the growing number of annotated protein spectra and the development of refined and standardized methodologies of protein extraction methods, matrix assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (MS) has emerged as a powerful technology for the prompt identification of microorganisms in clinical microbiology supporting genus-, or species-oriented treatment [31][32][33].
The DNA sequencing methods, targeting the 18S rRNA genes or ITS regions, can generate accurate species-level identification for many microbial isolates; however these methods are time consuming and technically demanding.
Polymerase chain reaction (PCR)-based techniques appear to be promising in terms of speed, economy, and resolution power [34]. PCR based patient follow up were shown to be capable to precede clinical signs of invasive candidiasis (IC) with the range of 1 day to 4 weeks, furthermore treatment could be initiated 3 days (range: 0-8 days) before the blood culture diagnosis was taken [35]. There have been numerous platforms and gene targets used for Candida spp. identification, such as those genes encoding cytochrome P450, actin and L1A1, and the highly variable internal transcribed spacer regions; ITS1 and ITS2 in combination with the relatively conserved regions of 18S, 5.8S or 28S nuclear rRNA genes [36][37][38][39][40][41].
MALDI-TOF and real-time PCR applications have high throughput, low-cost in supplies, short turnaround time. Both tools have the advantage to identify a broad-range of clinically relevant pathogenes with high accuracy at the species level Though MALDI-TOF is not suitable to detect a low amount of microorganisms directly from blood and PCR techniques usually lack standardization.
The prime aim of this study is to develop and evaluate a real-time PCR method which can identify and differentiate among seven relevant Candida spp. with high accuracy in a single, closed tube system, by a post-PCR T m calling assay coupled with a contingent high-resolution melting (HRM) analysis. This barcoding method is a single locus based HRM system which relies on the conserved regions of Candida beta-tubulin genes permitting reliable and precise species-specific identification of seven reference strains and 38 clinical isolates. To our knowledge, there is no other published method which is tailored to a single domain of Candida beta-tubulin genes and there is no other conventionally available or presented simplex PCR assay.

Fungal and bacterial strains
Genomic DNA (gDNA) samples of 81 clinically relevant fungal strains; Candida (43), aspergilli (32), Fusarium (4), Lichtemia (1), Rhyzopus (1), Scedosporium (1) and 16 bacteria; Gram-positive (10) and Gram-negative (6) were examined. The reference strains and clinical isolates (Table 1) were maintained at the Department of Microbiology, University of Szeged and at the Department of Medical Microbiology, University of Debrecen. To preserve the viability and purity of fungal and bacterial strains they were maintained in 50% glycerol stock solution at − 80°C and were periodically subcultured.

DNA extraction
All genomic DNA (gDNA) extraction steps were performed in a class II laminar-flow cabinet to avoid environmental contamination.

Fungal gDNA extraction
Candida cultures used for the molecular identification of the strains were grown on yeast-peptone D-glucose

Bacterial gDNA extraction
The bacterial strains were grown on Müller-Hinton agar base under aerobic conditions. DNA was extracted using the E.Z.N.A.® Bacterial DNA Kit (Omega Biotech, Norcross, Georgia, USA) per the manufacturer's instructions. Briefly, for suspension cultures one millilitre of log phase culture with approximately 10 8 bacterial cells was used and cells were pelleted by centrifugation. Gram-positive bacterial cell walls were lysed by lysozyme solution with 20 mg ml − 1 enzyme in 20 mM Tris HCL, pH 8.0 2 mM EDTA, 1.2% TritonX100). DNA concentrations and purity were measured using NanoDrop-1000 spectrophotometer (NanoDrop Technologies, Inc., North Carolina, USA).

CanTub-simplex PCR
Annotated sequences of the Candida beta-tubulin genes were extracted from the EMBL/GeneBank databases to make multiple alignments using Clustal Omega. Available Candida beta-tubulin sequences were aligned surveying potential sequence deviations within species. When designing primers, we screened for melting domains covering enough mismatches to enable proper discrimination among the tested strains but comprising 30-40 nucleotide long conserved nucleotides on flanking regions for the very specific hybridization of primers and avoiding cross-reactions with gDNA of other BSI causing fungal, bacterial strains and with the human genomic DNA.
Setting the optimal CanTub-simplex PCR conditions Temperature gradient assay was performed from 55 to 72°C for assessing the performance of the primer pair during amplification with a temperature gradient program using the LightCycler® 96 thermal cycler Instrument (Roche Applied Science, Penzberg, Germany). The MgCl 2 optimization was performed by adding different amounts of MgCl 2 within the concentration range 1 to 3.5 mM. The primer optimization assay was performed using 0.2, 0.4, 0.6 μM of the forward and reverse primers.

Verification of the CanTub-simplex amplicons via conventional PCR
In order to monitor the accumulation of aspecific amplicons in the CanTub-simplex PCR, primers were tested with 20 ng gDNA of five Candida reference strains (C. albicans ATCC 10231, C. glabrata ATCC 90030, C. parapsilosis ATCC 22019, C. tropicalis ATCC 750, C. krusei ATCC 6258), and two clinical isolates (C. guilliermondii SZMC 1536 and C. dubliniensis SZMC 1470 depicted by ID33 and ID36 in Table 1). The thermocycling reactions were conducted in a Roche LightCycler®Nano instrument (P2). 200 ng of the yielded CanTub PCR amplicons were electrophoresed on 2% TBE-agarose gel stained with ethidium-bromide to investigate aspecific PCR increments.

Real time PCR platforms
Real-time PCR assays were conducted on three different PCR platforms ( Temperature parameters were set as follows: an initial denaturing step of 95°C for 10 min followed by 50 cycles of denaturation at 95°C for 10 s, annealing at 60°C for 10 s and extension at 72°C for 10 s. Fluorescent data were collected in the ResolightDye channel (470/514 nm).

Melting temperature calculation and HRM protocols
The accumulation of the Candida species descriptive amplicons was monitored via melting temperature calculation and HRM analysis on three real-time PCR platforms. (P1), (P2) Following the completion of real-time PCR, the products were denatured at 95°C for 15 s (4.4°C s − 1 ), and then renatured at 40°C for 15 s (1°C s − 1 ) to form DNA duplexes. HRM analysis was performed by increasing the temperatures from 45 to 97°C (0.2°C s − 1 ) recording changes in fluorescence (-dF/dT) and plotting against changes in temperature. The HRM profiles were then analyzed using the LightCycler®96 HRM analysis and the Light-Cycler® Nano Software thus highly dense measurement points were taken during the high resolution melting stage resulting in species descriptive HRM melting curves.
(P3) HRM analysis was not performed with the LC 2.0 software due to the lack of this option. For T m analysis melting temperature measurement was set from 45 to 97°C (0.05°C s − 1 ). Melting peaks were analyzed by the LightCycler®2.0 Software.  Table 1.
Verification of the CanTub-simplex assay on EDTA-whole blood panels Obtaining of EDTA-WB samples for spike in controls Working with WB samples from healthy volunteers was approved by the local ethics committee, MK-JA/50/0096-01/ 2017. Participants agreed to take part in this study. EDTA whole blood samples obtained from healthy volunteers were pooled and screened for Candida contamination prior to use as extraction negative controls (ENC).

DNA purification from EDTA-WB samples
DNA purification steps were performed in a class II laminar air-flow cabinet to avoid environmental contamination.
Spiked EDTA-WB samples were disrupted (2000 rpm, 1 min) by Roche MagNa Lyser (Roche Diagnostics, Risch-Rotkreuz, Switzerland), and DNA was extracted along with the ENCs using the High Pure Viral Nucleic Acid Large Volume Kit (Roche Applied Science) according to manufacturer's instructions. To obtain technical duplicates every sample extractions were performed in parallel. The elution volumes were adjusted to 15 μl and pooled.

PCR efficiency and the limit of reliable detection (LoRD)
Amplification reactions were carried out on three different PCR platforms; P1, P2, P3. Duplicate PCRs were performed at every dilution of the Candida EDTA-WB reference panel extracts. Quantification cycle (Cq) values were subtracted and melting curves were analyzed to estimate the lowest template DNA concentration by which the appropriate, species descriptive melting peaks were interpretable (LoRD). To study the correlation between the Cq-s and the genomic load standard curves were obtained by plotting Cq values against the log of cell number; 10 6 -10 CFU/1 ml WB which is equivalent to 2 × 10 5 -0.2 GE/PCR. Mean Cq data were calculated and standard curves were built where these plots determined the linear dynamic ranges. Efficiency was calculated per the following formula, E = (10-1/slope) than was converted to percentage efficiency by using the formula, E% = (E-1) × 100 [44]. Amplification efficiency, (correlation coefficient; R 2 ) was also calculated.

In-house quality assessment
PCRs were conducted on P1 and T m data were subtracted.

Determining the repeatability
To estimate repeatability intra-assay consistency was calculated. To calculate the coefficient of variation (% C.V.) of the CanTub-simplex PCR within the three PCR plates (ref-plate_1, _2, _3) of the seven Candida reference panel extracts (ref-panel_1-_7) and within the four PCR plates (clin-plate_1, _2, _3, _4) of the seven Candida clinical panel extracts (clin-panel_1-_7) standard deviation (±SD) of duplicate T m mean data were taken, dividing these numbers by the mean of the T m values and multiplying them by 100 [see Additional file 1]. Finally, the grand mean of the sample coefficient of variations (average % C.V.-s) of the three plates was taken defining the plate % C.V. values. When the intra-assay % C.V. is below 10% the method under investigation has high precision.

Determining the reproducibility
To estimate the precision of the CanTub-simplex PCR as regards discrimination of the relevant Candida species and to confirm that results generated are consistent over time inter-assay consistency (plate-to-plate variation) was estimated between the PCR plates on three distinct days with the seven Candida reference panel extracts and with the sample extracts of the seven Candida clinical panels [see Additional file 2]. In the case of the Candida panels duplicate PCRs were performed on every sample isolate. T m data were subtracted than mean T m (°C) with standard deviations (±SD) were calculated. When analyzing Candida clinical panels, duplicate T m means of adherent clinical strains of the different species were assembled and overall mean was calculated. Plate coefficient of variations (% C.V.) was calculated. Finally, grand mean of the sample coefficient of variations (average % C.V.-s) was taken. Inter-assay % C.V. values less than 15% are generally acceptable.

In silico assessment of the CanTub forward and CanTub reverse primers
Based on the sequence alignment data we presumed that the amplicons generated by CanTub-simplex primers will display enough sequence divergence between closely related Candida species but at the same time may be conserved enough barcoding the different clinical isolates within species. Primer matches with annotated beta-tubulin genes from C. albicans, C. tropicalis, C. dubliniensis and C. glabrata are indicated in [see Additional file 3] with the sizes of the amplicons.

Optimal reaction conditions with the CanTub-simplex PCR
Optimal reaction conditions were determined as described in the methods section. 0.4 μM primers-, 2 mM MgCl 2 -concentrations and annealing at 60°C proved to be optimal. The size of the amplified CanTub-simplex PCR bands were verified on five Candida reference strains and on two clinical isolates. All the amplified CanTub-simplex PCR fragments are smaller than 300 bp in size and we did not experience any aspecific PCR increments (data not shown).

Cross reactivity
No cross-amplification was detected with the human genomic DNA and with the Gram-negative and Gram-positive bacteria. Furthermore, no cross-amplification was observed with A. terreus, A. lentulus, A. tubingensis, A. viridinutans and A. udagawae strains. Mild cross reaction was detected with the A. fumigatus, A. flavus and A. niger providing T m peaks at 86.35 ± 0.16°C and displaying inconclusive HRM curves and peaks. Moderate false positivity with plain, inconsistent peaks were detected in the case of Rhyzopus oryzae (77.38°C), Scedosporium aurantiacum (75.73°C) generating cycle threshold (Cq) values greater than Cq > 42 (data not shown).

Analytical sensitivity
LoRD of the simplex CanTub-PCR assay was estimated on Candida EDTA-WB standard panel extracts containing serially diluted Candida gDNA in a 6-log range. Linear dynamic ranges were assessed on P1, P2, and P3 with PCR efficiency (E) and coefficient of correlation (R 2 ) were also estimated using the standard curve data (Fig. 1). The analytical sensitivity with the lowest concentration of template DNA where reliable identifications with conclusive melting peaks and HRM curves were attainable in the case of at least one technical duplicate proved to be 0.2 GE/PCR (10 cells/ 1 ml WB) on all panels and on all platforms except the C. parapsilosis reference panel on P1 and P3, where LoRD proved to be 2 GE/PCR (100 cells/1 ml WB) only. On P2 the LoRD was 2 GE/PCR also for C. parapsilosis, C. dubliniensis and C. guilliermondii reference panels.

Maximum discriminatory property was achieved by defining seven melting clusters on three real-time platforms
CanTub-simplex assay targeting the beta-tubulin genes proved to be accurate and reliable for specific identification of seven Candida species. Thirty eight clinical isolates phenotypically identified as Candida species were subsequently tested in a blinded manner and barcoded due to comparing the melting data to those of the reference strains.
Normalized melting curves ( Fig. 2/a) with the difference plot ( Fig. 2/b) are shown by using saturating concentrations of a fluorescent double stranded DNA intercalating Reso-Light® dye. HRM analyses were performed on Candida EDTA-WB clinical panels and on platform P1.
Whisker plots show the distribution of the T m values when testing the 38 Candida clinical strains on four separate plates. For every dataset, the median, minimum, maximum T m values along with the 25th and 75th percentile are also shown in Fig. 3. Based on the distribution of the melting temperature data of the Candida strains tested, we created seven melting clusters for the reliable species level identification on P1-P3 C. dubliniensis cluster_1 T m with 95% CI: 77.58 ± 0.32°C (  unknown Candida strains there is no need for using standards as reference controls.

Repeatability and reproducibility
We estimated the intra-, and inter-assay consistency of the CanTub-simplex PCR on Candida EDTA-WB reference-[see Additional file 1] and on Candida EDTA-WB clinical panels [see Additional file 2]. For measuring the precision of the assay average coefficient of variation was calculated for the technical duplicate T m values measured on P1 where intra-assay % C. V. proved to be 0.14 (ref. plate_1), 0.19 (ref. plate_2) and 0.08 (ref. plate_3) on Candida EDTA-WB reference plates and 0.14 (clin. plate_1), 0.12 (clin. plate_2), 0.10 (clin. plate_3), 0.13 (clin. plate_4) on Candida EDTA-WB clinical plates accounting for a very high accuracy of the CanTub-simplex PCR. Plate-to-plate consistency was also assessed for the assay between the three Candida EDTA-WB reference plates and the four Candida EDTA-WB clinical panels. Reproducibility was measured and sample coefficient of variations (grand % C.V.-s) between the three reference and four clinical plates were taken, where inter-assay proved to be 0.11between standard [see Additional file 1] and 0.12between clinical panels [see Additional file 2] which is highly acceptable.

Discussion
With the continuously growing population of immunocompromised patients, the number of invasive fungal infections has increased significantly over the past decades [1][2][3][4]. Among all, Candida species are the most prevalent etiology of sepsis and septic shock in critically ill patient groups representing a significant health challenge with increasing medical and economic importance [7][8][9]. Concern is rising about the growing incidence of non-albicans infections and the emergence of their various intrinsic or acquired resistances. Over the last two decades, the occurrence of non-albicans species have emerged [12,13,17,18,26]. More than 98% of candidemia cases are caused by C. albicans, C. glabrata, C. parapsilosis, C. tropicalis, C. krusei, including further species such as C. dubliniensis, and C. guillermondii [14][15][16].
Identification of Candida species is important due to the differences in the antifungal susceptibility profile associated with species and because of the limitation of the phenotypic identification [10][11][12][13][14][15][16][17].
The traditional culture based morphological (colony and microscopic morphology), and/or biochemical (such as sugar assimilation and fermentation tests) identification methods of the Candida species are laborious, requiring a high level of skills and expertise of clinical mycologists and a long period of time, limiting patient care [27,28]. Morphological identification of certain species remains problematic due to the high degree of phenotypic similarities between C. albicans versus C. tropicalis and C. dubliniensis out of which the latter species can acquire stable fluconazole resistance rapidly [28]. Fig. 3 Whisker plots of the Candida species descriptive clusters. Whisker plots representing the discriminatory power of the CanTub-simplex PCR were created based on the digitalized melting temperature (T m ) data representing the seven melting clusters (pink cluster_1 for Candida dubliniensis, orange cluster_2 for Candida tropicalis, blue cluster_3 for Candida albicans, green cluster_4 for Candida krusei, yellow cluster_5 for Candida parapsilosis, brown cluster_6 for Candida guilliermondii, red cluster_7 for Candida glabrata) for the discrimination of the Candida species. Melting peaks were subtracted from the analysis of purified the Candida EDTA-WB clinical panel extracts (P1-P3). T m data of adherent clinical strains of the different species were assembled and data were grouped. Whisker plots were constructed for each data group showing the range of obtained temperatures of melting (T m ); the minimum, the median, and the maximum T m values with 25th and 75th percentiles There is an increasing demand for innovative sensitive, rapid and non-invasive methods for identification of Candida species at an early stage of the disease.
T m calling assays especially when coupled with HRM analysis has been introduced for scanning genotypes and for the rapid discrimination between DNA sequences based on their variant high resolution melting profiles without the use of fluorescent labeled probes. Although HRM based methods do not have the resolving power that many sequencing techniques have and are not as sensitive as the TaqMan probe based systems, they became more and more attractive to molecular diagnostic laboratories. Our recent data also support that applications resting on HRM analyses may be ideally suited for identifying or ruling out certain fungal and bacterial pathogens in clinical diagnostic workflow.
In 2007 Carvalho and colleagues developed and tested on 231 Candida isolates a sensitive (2.15 ± 0.25 CFU/ 1 ml), multiplex PCR based method by using yeast specific (ITS1, ITS2) universal and species-specific (18S, 28S) primers allowing the barcoding of eight relevant Candida strains (C. albicans, C. glabrata, C. parapsilosis, C. tropicalis, C. krusei, C. guilliermondii, C. lusitaniae and C. dubliniensis) at the species level taking the advantage of the presence of high-copy number of rRNA genes due to the presence of characteristic PCR amplified band patterns [36]. The only drawback of this method is that barcoding of unknown yeasts require running eight simultaneous PCR assays.
The CanTub-simplex PCR assay described here relies on a single primer pair targeting specific regions of Candida beta-tubulin genes.
The assay provides reliable nucleic acid based testing for proper identification of seven relevant Candida species by defining the species-specific melting domains and/or the shape of the derivative melting curves on the three real-time platforms (P1-P3).
When analyzing 1 ml of WB samples infected with 10 6 -10 Candida cells the amplification efficiency was 100% on all reference panels (ref. panel_1ref. panel_7) observing efficient amplification and melt curve analysis. For samples seeded with 1 CFU sample the barcoding capacity proved to be only 78.57%. Nevertheless, on all Candida reference panels the amplification efficiencies were reliable (E% = 1.85-1.94). It must be also noted, that the standard deviations (mean SD: ±0.12°C) of the T m data measured on reference panels proved to be low even in bright concentration ranges.
Here we demonstrated that when appropriate DNA samples are available (OD 260/280 1.82-1.97 and gDNA concentration of 1-10 ng/μl), CanTub-simplex PCR assay identifies the seven most frequent pathogenic Candida species. CanTub-simplex PCR based applications used in parallel with morphotyping may offer better resolution of the species level identification. This method is technically not demanding and provides clear, easily traceable protocols on three different real-time platforms. Single locus primer pair is used so the method can be easily multiplexed with exogenous or endogenous TaqMan internal control assays. Here we also demonstrated that the adaptation of DNA based applications used in tandem with morphological examinations in clinical diagnostic laboratories can offer better resolution of species within the genus.

Conclusion
CanTub-simplex PCR targeting the beta-tubulin genes proved to be sensitive and accurate on all platforms (P1 -P3) tested here, and has the potential to identify seven clinically relevant Candida species (C. albicans, C. glabrata, C. parapsilosis, C. tropicalis, C. krusei, C. guilliermondii, C. dubliniensis) by using the seven, pre-defined melting clusters without using reference controls. These features make this CanTub-simplex PCR advantageous for use as a first-pass diagnostic adjunct in microbiology laboratories on blood culture bottles or for direct whole blood testing by making the patient follow-up more achievable.

Additional files
Additional file 1: Table S1. Intra-, and inter-assay coefficient of variation of the CanTub-simplex PCR on Candida reference panels. Representation of the technical duplicates of the T m data with mean and ± SD measured on Candida EDTA-WB reference panels at every dilution in a 6-log range. Intra-assay coefficient of variation was estimated for the particular gDNA panels on separarte plates (% C.V.-s of Candida ref-panel_1-7) even as for the plates in a whole (% C.V.-s of ref-plate_1-3). Inter-assay coefficient of variation was estimated for the CanTub-simplex PCR for the reference samples of all panels at every dilution in a 6-log range between three plates finally, grand mean of the sample coefficient of variation (grand % C.V. on ref-plates) was calculated. (DOCX 37 kb) Additional file 2: Table S2. Intra-assay coefficient of variation of the CanTub-simplex PCR on 38 Candida clinical strains via EDTA-WB Candida clinical panel. T m data of Candida clinical strains were grouped according to panels (clin-panel_1-7) and T m mean was calculated with ±SD. Coefficient of variation (% C.V.) was calculated for every species. Finally, intra-and inter-assay consistencies were calculated. (DOCX 21 kb) Additional file 3: Figure S3