THE COMPARISON OF THREE REAL-TIME PCR KITS FOR SARS-COV-2 DIAGNOSIS REVEALS DISCREPANCIES ON THE IDENTIFICATION OF POSITIVE COVID-19 CASES AND DISPERSION ON THE VALUES OBTAINED FOR THE DETECTION OF SARS-COV-2 VARIANTS

The COVID-19 pandemic has generated a huge challenge and threat to public health throughout the world population. Reverse transcription associated with real-time Polymerase Chain Reaction (RT-qPCR) has been the gold-standard molecular tool for diagnosis and detection of the SARS-CoV-2. Currently, it is used as the main strategy for testing, traceability, and control of positive cases For this reason, the on-top high demand for reagents has produced stock-out on several occasions and the only alternative to keep population diagnosis has been the use of different RT-qPCR kits. Therefore, we evaluate the performance of three of the commercial RT-qPCR kits currently in use for SARS-CoV-2 diagnosis in Chile, consisting in: TaqMan 2019-nCoV Assay Kit v1 (Thermo). Real-Time Fluorescent RT-PCR Kit for Detecting SARS-CoV-2 (BGI), and LightCycler Multiplex RNA Virus Master (Roche). Results of quantification cycle (Cq) and relative fluorescence units (RFU) obtained from their RT-qPCR reactions revealed important discrepancies on the total RNA required for the identification of SARS-CoV-2 genes and diagnosis. Marked differences between kits in samples with 30>Cq value< 34 was observed. Samples with positive diagnoses for Covid-19 using the Thermo Fisher kit had different results when the same samples were evaluated with Roche and BGI kits. The displacement on the Cq value for SARS-CoV-2 identification between the three different RT-qPCR kits was also evident when the presence of single nucleotide variants was evaluated in the context of genomic surveillance. Taken together, this study emphasizes the special care adjusting RT-qPCR reaction conditions of the different kits must be taken by all the laboratories before carrying out the detection of SARS-CoV-2 genes from total RNA nasopharyngeal swab (NPS) samples.


INTRODUCTION 4 6
Coronavirus disease 2019  is caused by the severe acute respiratory syndrome 4 β -actin were included and assessed individually in 1 2 1 the 96-well PCR plate. The polymerase from BGI Reaction Mix (BGI Health (HK) Co. Ltd, China, 1 2 2 Cat. No. MFG030010) was included in each reaction. Each reaction contained 18.5 µl of SARS-1 2 3 CoV-2 Reaction Mix (HEX detector channel to β -actin and FAM detector channel to ORF1ab), 1.5 1 2 4 µl SARS-CoV-2 Enzyme Mix, 8 µl of nuclease-free water, and 2 µl of extracted RNA sample. 1 2 5 When 10 µl of extracted RNA was used as template, nuclease-free water was not dispensed in the 1 2 6 reaction. The amplification thermal conditions include the reverse transcription at 50 °C for 5 1 2 7 minutes, predenaturation at 95 °C for 20 s, followed by 45 cycles at 95 °C for 3 seconds and 60 °C 1 2 8 for 30 seconds. The LightCycler ® Multiplex RNA Virus Master kit detects viral SARS-COV-2 1 2 9 genome sequence using the RdRP probe (LightMix ® Modular Wuhan CoV RdRP-gene. Cat. No. 1 3 0 53-0777-96) using a one-step strategy. Positive internal control probe for RdRP (LightMix ® 1 3 1 Modular Wuhan CoV RdRP-gene. Cat. No. 53-0777-96) was included and assessed individually in 1 3 2 the 96-well PCR plate. As reference control, the RNase P probe (TaqMan™ 2019-nCoV Control 1 3 3 Kit v1; Thermo Fisher Scientific, Cat. No. A47533) was included for ensuring the presence of total 1 3 4 RNA extracted from NPS samples as template. This decision was supported on the antecedent the 1 3 5 Roche RT-qPCR kit utilized the Equine Arteritis Virus (EAV) as an internal control for the 1 3 6 extraction process but not a control of the total RNA extracted. The polymerase from RT-qPCR 1 3 7 Reaction Mix 5x (The LightCycler ® Multiplex RNA Virus Master kit, Cat. No. 06754155001) was 1 3 8 included in each reaction. Each reaction contained 0.5 µl of RdRP (FAM detector channel), 4µl of 1 3 9 RT-qPCR Reaction Mix 5x, 0.1 µL of RT Enzyme Solution 200x, 1 µL of RNase P probe, 12.4 µl 1 4 0 of nuclease-free water, and 2 µl of extracted RNA sample. When 5 µl of extracted RNA was used 1 4 1 as template, 9.4 µl of nuclease-free water were dispensed in the reaction. The amplification thermal 1 4 2 conditions include the reverse transcription at 50 °C for 10 minutes, predenaturation at 95 °C for 30 1 4 3 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2021. To establish PCR efficiency and the detection limit for both the reference (RNase P for Thermo 1 4 9 Fisher and Roche RT-qPCR kits; beta-actin for BGI kit) and the viral genes assessed (ORF1ab for 1 5 0 Thermo Fisher and BGI kits; RdRP for Roche kit) we ran RT-qPCR reactions using serial dilutions. 1 5 1 In order to get the maximum representation of values in the curve, we used for the 10-fold serial 1 5 2 dilutions a reference pool made from randomized ten total RNA NPS-extracted samples with a Cq 1 5 3 value around 20. The reactions were carried out in triplicate according to the specific conditions 1 5 4 indicated by the manufacturer and described above. All the RT-qPCR reactions were performed on 1 5 5 the Agilent AriaMx Real-Time PCR System. We determined the slope by linear regression in 1 5 6 GraphPad Prism and defined the required levels for PCR efficiency (E) and R-squared (R 2 ) as>95 1 5 7 % and>0.95, respectively. The primer efficiency was calculated according to the formula Efficiency 1 5 8 % (E) = (10 (-1/Slope)-1) ) x 100 [8].
To determine an approach about the detection limit we select ten 1 5 9 samples with Cq values above 30. To determine the minimum detection limit for each RT-qPCR 1 6 0 SARS-CoV-2 detection kit, a standard curve for the amplification of each probe assessed was 1 6 1 generated. The detection limit was established based on the last dilution on all the triplicates 1 6 2 amplified. We also took into consideration the R 2 (intended as a goodness-of-fit measure for linear 1 6 3 regression) and the probe efficiency (closer to 100%, intended 100% as the optimum probe 1 6 4 efficiency value). positive patients were added. The thermal profile consists of a reverse transcription phase for 15 1 7 5 minutes at 50 ° C and an activation phase at 95 ° C for 5 minutes. Then, for PCR reaction 45 1 7 6 amplification cycles were run with a denaturation phase for 5 seconds at 95ºC, an annealing / 1 7 7 extension phase for 30 seconds at 57ºC and a scan phase within each cycle for the different probes. 1 7 8 The data obtained was exported in an Excel spreadsheet and the Cq value and fluorescence relative 1 7 9 intensity was analyzed for the internal positive control, IPC (TAMRA) and each one of the variants 1 8 0 assessed. Government of Chile) to which they attended on their own. Verbal consent was detailed by the 1 8 8 health professional assigned by CESFAM for this purpose. Once their consent was given, the 1 8 9 patient gave their data to the health professional to identify, trace, and isolate a possible positive 1 9 0 case of Covid-19. Once the sample was received in the diagnostic laboratory, the person in charge 1 9 1 of the sample reception team (Dr. Claudio Acuña-Castillo) assigned an internal sample code to 1 9 2 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 16, 2021. ensure the traceability of the sample. Thus, data analysis used for this study was conducted only 1 9 3 using the internal sample code numbers assigned at the moment to receive the nasopharyngeal swab 1 9 4 samples for diagnostics purpose. Accordingly, the samples have been irreversible anonymized for 1 9 5 analysis and interpretation of results by the diagnostic laboratory team. Once assigned the 1 9 6 diagnostic result for each sample, Dr. Acuña-Castillo was responsible for communicating the result 1 9 7 to the CESFAM of origin for each sample. 1 9 8 2.7 Data representation and statistical analysis. A paired two-sided Student T-test was used to 1 9 9 determine differences between the Cq and RFU obtained from the different SARS-CoV-2 RT-qPCR 2 0 0 detection kits. A p-value of < 0.05 was considered statistically significant. GraphPad Prism 8 2 0 1 statistical software was used to analyze and plot the data obtained.
The analysis of the extracted NPS samples with the Thermo Fisher RT-qPCR kit using 5 µl 2 0 5 (according to the manufacturer instructions) and 2 µl of total RNA revealed important differences 2 0 6 both in the quantification cycle (Cq) and in the relative fluorescence units (RFU) determined by the 2 0 7 RNase P (reference gene) and ORF1ab (SARS-CoV-2 gene) amplification (Fig 1). From a global 2 0 8 perspective, the 2 µl of total RNA template decreased the Cq value in most of the samples assessed 2 0 9 compared to the 5 µl of total RNA template ( Fig 1A). This first perception is reinforced when the 2 1 0 mean ± SD is represented, showing a lower Cq mean value for the 2 µl of total RNA template 2 1 1 (14.09 ± 0.99) than the 5 µl of total RNA template (14.82 ± 0.98) (Fig 1B). The same behavior was 2 1 2 also observed for the RFU, registering a strong difference between both volume of templates ( Fig  2  1 3 1C) and determined by a higher mean fluorescence for the 2 µl of total RNA template (8969 ± 2 1 4 1232) than the 5 µl of total RNA template (4041 ± 981) ( Fig 1D). When the presence of SARS-2 1 5 CoV-2 genome was evaluated by RT-qPCR in the total RNA extracted from NPS samples, those 2 1 6 three samples diagnosed as COVID-19 positive using 5 µl of total RNA showed quite similar Cq 2 1 7 values using 2 µl of total RNA as template (from lower to higher Cq value: Cq 5ul = 21.96 and 2 1 8 Cq 2ul =21.10; Cq 5ul = 35.13 and Cq 2ul = 36.69; Cq 5ul = 36.15 and Cq 2ul = 36.53) ( Fig 1E). However, 2 1 9 other three total RNA NPS-extracted samples diagnosed as COVID-19 negative using the 5 µl of 2 2 0 total RNA was diagnosed as COVID-19 positive with a template of 2 µl of total RNA (Cq 5ul = 46.00 2 2 1 and Cq 2ul = 35.18; Cq 5ul = 46.00 and Cq 2ul = 36.92; Cq 5ul = 39.99 and Cq 2ul = 37.03). Based on these 2 2 2 results, it is not a surprise that the Cq mean for the 2 µl of total RNA template (36.39 ± 5.02) was 2 2 3 lower than the 5 µl of total RNA template (40.44 ± 7.29) ( Fig 1F). In the same way than it was 2 2 4 observed for the amplification of the RNase P reference gene, all the total RNA NPS-extracted 2 2 5 samples registered a much higher fluorescence for the 2 µl compared to the 5 µl of total RNA as 2 2 6 template ( Fig 1G). Thus, the 2 µl of total RNA triplicated its mean fluorescence value (1705 ± 2 2 7 1553) in comparison with the 5 µl of total RNA (544.6 ± 562.3) (Fig 1H). The results with Thermo 2 2 8 Fisher RT-qPCR kit suggest a higher sensitivity of SARS-CoV-2 using 2 µl of total RNA instead 2 2 9 the 5 µl recommended by the manufacturer.
The BGI RT-qPCR kit registered also differences between the volume recommended by the 2 3 2 manufacturer (10 µl) and 2 µl of total RNA. At first sight, the amplification of beta-actin (internal 2 3 3 control) showed apparently a slight lower Cq values for the 10 µl of total RNA template in most of 2 3 4 the samples assessed compared to the 2 µl of total RNA template (Fig 2A). This data is supported 2 3 5 by the mean ± SD of all analyzed samples, effectively showing a slight decrease on the Cq mean 2 3 6 value for the 10 µl of total RNA template (23.21 ± 1.25) than the 2 µl of total RNA template (23.58 2 3 7 ± 1.20) (Fig 2B). By contrast, the opposite trend observed for Cq values was observed for RFU, 2 3 8 noting an apparent higher fluorescence when it was used a volume of 2 µl of total RNA as template 2 3 9 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 16, 2021. ; https://doi.org/10.1101/2021.07.13.21260484 doi: medRxiv preprint ( Fig 2C). This perception is confirmed by the higher RFU mean using 2 µl (4216 ± 698.5) than the 2 4 0 10 µl of total RNA template (3724 ± 860.5) (Fig 2D). When the presence of SARS-CoV-2 was 2 4 1 evaluated in the total RNA NPS-extracted samples, very similar results were observed between the 2 4 2 paired Cq values for 10 and 2 µl of total RNA (Fig 2E). However, four samples diagnosed as 2 4 3 COVID-19 negative with 10 µl of total RNA were determined as COVID-19 positive when 2 µl of 2 4 4 total RNA were dispensed (Cq 10ul = 39.66 and Cq 2ul =31.77; Cq 10ul = 46.00 and Cq 2ul = 35.05; Cq 10ul = 2 4 5 46.00 and Cq 2ul = 33.54; Cq 10ul = 46.00 and Cq 2ul = 28.79) (Fig 2F). This result is influencing upon a 2 4 6 slight lower Cq mean value for SARS-CoV-2 ORF1ab detection with 2µl (27.70 ± 7.16) than 10 µl 2 4 7 (29.64 ±9.52). In the same way than it was observed for the amplification of the Beta-actin 2 4 8 reference gene, all the total RNA NPS-extracted samples registered a much higher fluorescence for 2 4 To evaluate the performance of the three RT-qPCR, we compared the Cq, RFU and COVID-19 2 7 8 diagnosis on 90 randomized total RNA NPS-extracted samples. The amplification of the reference 2 7 9 gene showed clear differences between the RT-qPCR kits assessed, showing a greater Cq value on 2 8 0 most of the samples evaluated with the Thermo Fisher kit (Fig 4A). By contrast, the samples 2 8 1 amplified with the BGI kit showed the lower Cq value, even identifying two samples behind the 2 8 2 beta-actin detection limit (Fig 4A). The differences observed for the RNase P paired data was 2 8 3 confirmed with the Cq mean value for each kit, noting the highest Cq mean value for the Thermo 2 8 4 Fisher kit (16.55 ± 2.37), followed by the Roche kit (18.28 ± 2.03), and the BGI kit (28.01 ± 3.01) 2 8 5 ( Fig 4B). The same profile was observed for the reference gene paired data amplification (Fig 4C), 2 8 6 noting the greatest and the lowest Cq mean values for the Thermo Fisher (7351 ± 1109) and the 2 8 7 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 16, 2021. ; https://doi.org/10.1101/2021.07.13.21260484 doi: medRxiv preprint BGI kits (2416 ± 482), respectively ( Fig 4D). Importantly, the amplification profile observed for the 2 8 8 reference gene was not the same for the SARS-CoV-2 gene amplification. The paired Cq data 2 8 9 showed the greater Cq value for the Thermo Fisher kit but now followed by the BGI instead the 2 9 0 Roche kit, although no significant differences were observed between the BGI and the Roche kit 2 9 1 because the similar Cq mean value for both kits (29.63 ±6.87; 29.94 ± 6.08) (Fig 4E). Importantly, 2 9 2 the differences between the Cq mean value on the viral gene for Thermo Fisher kit (27.24 ±5.65) 2 9 3 and the other two kits is probably the responsible of the discrepancies observed in the COVID-19 2 9 4 positive diagnosis for the samples evaluated (23 positive samples diagnosed by Thermo Fisher; 18 2 9 5 positive samples diagnosed by BGI; 17 positive samples diagnosed by Roche) (Fig 4I). At fluoresce 2 9 6 level, the paired and mean values showed the same trend than the Cq values, although highlighting 2 9 7 the statistical difference also observed between the BGI and Roche RT-qPCR kits (Fig4 G-H). 2 9 8 Based on these results, we hypothesize that the discrepancies observed between the RT-qPCR kits 2 9 9 evaluated were focused on total RNA NPS-extracted with a high Cq (Cq > 30). Thus, the 3 0 0 comparison for the SARS-CoV-2 diagnostic performance was evaluated between the Thermo 3 0 1 Fisher, BGI and Roche RT-qPCR kits using ten randomized NPS samples with low Cq value (19< 3 0 2 Cq value< 25 for the ORF1ab amplification using the Thermo Fisher RT-qPCR kit), and other ten 3 0 3 randomized NPS samples with high Cq value (30< Cq value< 34 for the ORF1ab amplification 3 0 4 using the Thermo Fisher RT-qPCR kit). The reference gene amplification showed differences for 3 0 5 the paired Cq values between the RT-qPCR kits both for those samples identified with low Cq value 3 0 6 and high Cq value (Fig 5A; Fig 5I). Both in the low and high Cq value sample cases, the reference 3 0 7 gene amplification was greater for the Thermo Fisher kit ( Fig 5J). The differences for the paired RFU between kits had the same trend than it was observed 3 1 0 for Cq values, although even more marked when the RT-qPCR kit were compared (Fig 5C; Fig 5K). 3 1 1 Thus, in those samples with low and high Cq values, the RFU was much greater in the case of 3 1 2 Thermo Fisher kit (7455 ± 734; 7431 ± 535), followed by the Roche kit (4381 ± 633; 5093 ± 695), 3 1 3 and the BGI kit (2154 ± 522; 1919 ± 228) ( Fig 5D; Fig 5L). Importantly, when the amplification of 3 1 4 the viral gene was evaluated, it was not observed the same trend registered for the reference gene 3 1 5 amplification. In fact, in the paired-comparison perspective all the RNA NPS-extracted samples 3 1 6 with low and high Cq values showed the greater Cq viral gene amplification for the Thermo Fisher 3 1 7 kit but now followed by the BGI and the Roche kit ( Fig 5E; Fig 5M). In this way, in the samples 3 1 8 with low Cq values, their mean values showed slight but significant differences between kits (22.03 3 1 9 ± 1.67 for Thermo Fisher; 23.69 ± 2.65 for BGI; 25.39 ± 3.66 for Roche) (Fig 5F). However, in the 3 2 0 case of the samples with high Cq values more marked differences between kits were registered 3 2 1 (31.98 ± 1.03 for Thermo Fisher; 34.27 ± 1.91 for BGI; 43.27 ± 3.66 for Roche) (Fig 5N). These 3 2 2 differences are probably attributable not only to the sensitivity of each kit but also to the number of 3 2 3 samples diagnosed as COVID-19 positive. Moreover, meanwhile in the case of samples with low 3 2 4 Cq value all the ten samples were reported with COVID-19 diagnosis (Fig 5Q), in the samples with 3 2 5 high Cq just all the ten samples were effectively confirmed with COVID-19 positive diagnosis by 3 2 6 Thermo Fisher but only seven and even no one sample were diagnosed using the BGI and Rocke 3 2 7 kits, respectively (Fig 5N; Fig 5R). The same behavior trend in the RFU was observed for the low 3 2 8 and high Cq value samples, both in the paired (Fig 5G; Fig 5O) and RFU mean value obtained (Fig  3  2  9 5H; Fig 5P), respectively. These results indicate that the samples with Cq values greater than 30 3 3 0 could compromise its COVID-19 diagnosis depending on the kit used for this purpose. To determine the distribution of positive samples and its impact on the detection of single 3 3 3 nucleotide polymorphisms (also called single nucleotide variants, SNV) associated to SARS-CoV-2 3 3 4 variants, we evaluated twelve total RNA samples with a Cq value lower than 26. As it was 3 3 5 expected, all the twelve samples were identified as SARS-CoV-2 positive samples. However, from 3 3 6 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 16, 2021.
the twelve samples only six of them were also positive for the variants N501Y, K417N/T, and 3 3 7 E484K, suggesting the presence of the P1 (Gamma) SARS-CoV-2 variant in the 50% of the 3 3 8 samples (Fig 6A). By contrast, none of the samples were positive for Hv 69/70 del, and/or P681H. 3 3 9 According to our previous evidence, the Cq mean ± standard deviation (mean ± SD) for these 3 4 0 samples also showed a clear dispersion on the Cq distribution ( Fig 6B). In fact, one of the samples 3 4 1 identified with the SARS-CoV-2 N501Y, K417N/T, and E484K SNV registered a Cq RdRP value= 3 4 2 29.98 with the Roche RT-qPCR diagnostic kit (Fig 6B). The same sample showed a Cq ORF1ab value= 3 4 3 25.71 and Cq ORF1ab value= 27.13 for Thermo Fisher and BGI, respectively ( Supplementary Fig 2).

4 4
This evidence suggests that the recommendation for using only samples with Cq< 30 made by the 3 4 5 manufacturer should has also in consideration the diagnostic RT-qPCR used for such purpose. 3 4 6 sample using the recommended volume of extracted RNA (10 µl of total RNA, recommended by 5 0 0 the manufacturer; red spots) and 2 µl of total RNA (blue spots). In the graphs, each spot is a 5 0 1 different analyzed sample for each volume condition (10 µl; 2 µl). The line linking two spots 5 0 2 indicated the paired result obtained for the same sample assessed using the two different volume ORF1ab gene amplification obtained by RT-qPCR from all the samples evaluated by RT-qPCR. For 5 1 5 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 16, 2021. ; https://doi.org/10.1101/2021.07.13.21260484 doi: medRxiv preprint statistical analysis, paired two-sided Student T-test was applied (n= 71 NPS samples chosen at 5 1 6 random). * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001. 5 1 7 5 1 8 5 1 9 5 2 0 5 2 1 Fig 3. Comparative analysis for the detection of SARS-CoV-2 from NPS samples using RNase 5 2 2 P (as cellular reference gene) and RdRP gene (Roche RT-qPCR kit). The comparison was made 5 2 3 from the same NPS sample using the recommended volume of extracted RNA (5 µl of total RNA, 5 2 4 recommended by the manufacturer; red spots) and 2 µl of total RNA (blue spots). In the graphs, 5 2 5 each spot is a different analyzed sample for each volume condition (5ul; 2 ul). The line linking two 5 2 6 spots indicated the paired result obtained for the same sample assessed using the two different 5 2 7 volume conditions. Samples with Cq= 46 denotes no amplification. (A) Paired quantification cycle 5 2 8 (Cq) analysis for the RNase P the amplification values obtained by RT-qPCR for each sample 5 2 9 assessed. (B) Cq mean ± standard deviation (mean ± SD) for the RNase P amplification values random using the three RT-qPCR kits. The comparison was made from the same NPS sample 5 4 4 using the optimized volume of total RNA extracted (2 µl). In the graphs, each spot for each RT-5 4 5 qPCR kit is a different analyzed sample. The line linking the spots indicated the paired result 5 4 6 obtained for the same sample assessed by the different RT-qPCR kits. Samples with Cq= 46 denotes 5 4 7 no amplification. (A) Paired quantification cycle (Cq) analysis for the RNase P amplification values 5 4 8 obtained by RT-qPCR for each sample assessed. (B) Cq mean ± standard deviation (mean ± SD) for 5 4 9 the RNase P amplification obtained by RT-qPCR for all the samples evaluated. On (A) and (B), the 5 5 0 horizontal red (for Thermo Fisher), blue (for BGI), and green line (for Roche) indicates the 5 5 1 detection limit for the determination of the reference gene on each of the RT-qPCR kits (determined 5 5 2 on Supplementary Figure 1). (C) Paired relative fluorescence unit (RFU) analysis for the RNase P 5 5 3 amplification values obtained by RT-qPCR for each sample assessed. (D) RFU mean ± standard 5 5 4 deviation (mean ± SD) for the RNase P amplification obtained by RT-qPCR from all the samples 5 5 5 evaluated. (E) Paired Cq analysis for the SARS-CoV-2 ORF1ab gene amplification values obtained 5 5 6 by RT-qPCR for each sample assessed. (F) Cq mean ± standard deviation (mean ± SD) for the 5 5 7 SARS-CoV-2 gene amplification obtained by each one of the RT-qPCR assessed. On (E) and (F), 5 5 8 the horizontal red (for Thermo Fisher), blue (for BGI), and green line (for Roche) indicates the 5 5 9 detection limit for the determination of the viral gene on each of the RT-qPCR kits (determined on 5 6 0 Supplementary Figure 1). (G) Paired RFU analysis for the SARS-CoV-2 ORF1ab gene 5 6 1