Engineered RNA biosensors enable ultrasensitive SARS-CoV-2 detection in a simple color and luminescence assay

  1. Anirudh Chakravarthy
  2. Anirudh Nandakumar
  3. Geen George
  4. Shyamsundar Ranganathan
  5. Suchitta Umashankar
  6. Nishan Shettigar
  7. Dasaradhi Palakodeti
  8. Akash Gulyani
  9. Arati Ramesh

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

The continued resurgence of the COVID-19 pandemic with multiple variants underlines the need for diagnostics that are adaptable to the virus. We have developed toehold RNA–based sensors across the SARS-CoV-2 genome for direct and ultrasensitive detection of the virus and its prominent variants. Here, isothermal amplification of a fragment of SARS-CoV-2 RNA coupled with activation of our biosensors leads to a conformational switch in the sensor. This leads to translation of a reporter protein, for example, LacZ or nano-lantern that is easily detected using color/luminescence. By optimizing RNA amplification and biosensor design, we have generated a highly sensitive diagnostic assay that is capable of detecting as low as 100 copies of viral RNA with development of bright color. This is easily visualized by the human eye and quantifiable using spectrophotometry. Finally, this PHAsed NASBA-Translation Optical Method (PHANTOM) using our engineered RNA biosensors efficiently detects viral RNA in patient samples. This work presents a powerful and universally accessible strategy for detecting COVID-19 and variants. This strategy is adaptable to further viral evolution and brings RNA bioengineering center-stage.

Article activity feed

  1. Version published to 10.26508/lsa.202101213
  2. Review Commons

    Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Reviewer #1 (Evidence, reproducibility and clarity (Required):

    In this project, authors develop a colorimetric and luminescence assay for the detection of SARS-CoV-2 RNA in vitro. They design an RNA based sensor that will be triggered by target RNA then release the ribosome binding site and a translation start site followed by a reporter gene. The released sequence will then trigger the production of reporter protein by transcription-translation coupled assay. Authors also introduce an RNA amplification step in order to increase the sensitivity of this assay.

    **Strengths:**

    This assay provides a simple, rapid way to detect SARS-CoV2 and it is an elegant way to incorporate transcription-translation coupled assay for SARS-CoV-2 RNA detection and identify SARS-CoV-2 patient samples. It is a nice assay and the performance is comparable with the existing method.

    **Weaknesses:**

    However, the positioning of this assay is not very clear. The readout of this assay could be recorded by camera whereas it includes several steps such as RNA extraction, amplification, transcription-translation coupled assay and reporter reaction. The limitations of the existing methods (RT-PCR, paper strip) and the advantages of this assay haven't been demonstrated by the experiments. The stability of RNA may also restrict the application of the proposed assay on site.

    **Major comments:**

    Authors are suggested to design an experiment to show the advantage of this assay compared with the existing method.

    __Response: __We thank the reviewer for pointing this out. In Fig 5, we show a comparison of our assay with the bench mark in COVID-19 diagnostics, which is the RT-qPCR assay. We specifically correlate the Ct- values obtained for RT-qPCRs with the amount of color or luminescence obtained through our assay. From these experiments we note that the sensitivity of our assay is a lttle less than the RT-qPCRs where our assay does not detect Ct-values in the 36 to 38 range (very low viral loads). This comparative experiment highlights that our assay bears clear advantages over the RT-qPCR in terms of ease of assay set up, ease of color detection, amenability to cell-phone imaging and no requirement of sophisticated equipment or technical training to interpret results. The full details of these comparisons are discussed in the manuscript.

    This is consistent with the literature on COVID-19 diagnostics where new assays are routinely bench-marked against the “gold-standard” RT-qPCR assay ((Corman et al., 2020; Pearson et al., 2021).

    What is the limit of detection of this assay using LacZ and Luciferase reporter respectively?

    __Response: __The limit of detection of the assay as shown in Fig 4B and Fig 4C-D, was found to be 100 copies of RNA, which translates to a concentration of 8 attomolar RNA. In this case, we find the limit of detection to be the same for both LacZ (Fig 4B) and Luciferase (Fig 4C-D) reporter.

    The calculations of copy number and sensitivity were made using a commercial source of synthetic CoV-2 RNA (Twist Biosciences) that is used in several studies about COVID-19 diagnostics (Joung et al., 2020; Rabe & Cepko, 2020; Wu et al., 2021). The RNA copy numbers are taken from the product details provided by the manufacturer. These details are now clearly stated in the manuscript. The commercial RNA is provided at 106 copies per ul. From this we take as low as 100 copies per 20ul of NASBA reaction, which we are able to detect using our assay. Hence our sensitivity comes to 8 attoMolar. We have clarified this in the manuscript. __We noticed a typo in the original submission where we refer to a sensitivity of 80 attomolar in the Discussion. This is corrected to 8 attomolar. __With this sensitivity we are within the range to detect RNA in patient samples, as confirmed by our patient data.

    Authors have not examined the selectivity of this assay. What is the specificity, selectivity for each of these variants? Does altering target RNA change the specificity?

    __Response: __We thank the reviewer for raising this point. As recommended by the reviewer, we have now examined the selectivity of this assay through new data (See new Fig S3, new Fig S4 and new Fig S8, also shown below).

    We have examined selectivity in 3 different ways.

    1. Is our sensor selective to the said region of the SARS-CoV-2 genome? To address this, we generated 19 different Target (Trigger) RNAs spread across the SARS-CoV-2 genome. These were tested against Sensor 12 to examine for their ability to trigger the sensor. We find that our sensor is highly selective for its target RNA and does not show any detectable response to the other regions of SARS-CoV-2 (see new Fig S3).

    Next, we asked if our assay is selective to SARS-CoV-2 versus other related human corona viruses. For this, we first examined the sequence of the target RNA (Amplicon RNA 12) that is sensed by Sensor 12. We selected equivalent regions of RNA from a different coronavirus, the HKU1 human coronavirus family. We generated these RNA sequences in vitro and performed IVTT. These new data are shown in new Fig S4 and below. We find that the human coronavirus (HKU1) RNAs are not able to turn on our sensor, whereas the cognate SARS-CoV-2 RNA is able to.

    We then asked if our assay can detect a current prominent variant of SARS-CoV-2. A major cause of concern is the ability of SARS-CoV-2 to accumulate mutations in its genome, resulting in different variant strains of SARS-CoV-2. Of these variants, the Delta variant (B.1.617.2) is not only highly contagious but has been noted as a possible vaccine breakthrough mutant of SARS-CoV-2. For this, we obtained RNA from the patient nasopharyngeal swab samples from the NCBS-inStem Covid-19 testing Center, Bangalore, India. RNA was isolated in the BSL-3 facility at the testing center. RNA samples were sequenced and confirmed to be the Delta variant- B.1.617.2 (sequences deposited in GASIAD). RNA extracted from these patient samples were tested against Sensor 12 using NASBA followed by IVTT. We find that our assay can efficiently detect the Delta variant SARS-CoV-2 RNA from patient samples with a build up of color, but no color was observed from control samples. These new data are shown below and in new Fig 5F and new Fig S8. The ability to detect the Delta variant of SARS-CoV-2 is an important feature of our sensor since this variant is now of global concern and extensively found in the population, even becoming the dominant variant in several countries (Callaway, 2021; O’Dowd, 2021; Torjesen, 2021).

    In Figure 2C-F, sensor 17 showed higher fold change and sensitivity. Why was sensor 12 selected for further study in Figure 3

    __Response: __The reviewer rightly notes that sensor 17 responds to 1012 copies of RNA and hence appears to be inherently more sensitive than sensor 12, which responds to 1013 copies of RNA. However, neither of these sensitivities are good enough to detect the levels of viral RNA found in patient samples. Hence we coupled these sensors with a step of NASBA amplification. The screen to identify pairs of NASBA primers gave us great hits for sensor 12 right off the bat, where we could detect down to 100 copies of RNA. Hence we moved forward with sensor 12 for further experiments. This has now been clarified in the manuscript.

    Authors should show the error bar in all plots. Authors should also indicate what the error bar means (SD, S.E.M. etc.) throughout the manuscript.

    __Response: __This is an important point. We have added the error bars and statistical analyses to all relevant plots. We have included the description of these statistical parameters in the figure legends throughout the manuscript, where relevant. Alternatively, experimental replicates are indicated and shown in the revised manuscript. Specifically in Figures 2 and 3 and 4D we have performed statistical analysis to include p-values to show significance of the data. For the data in Figure 4 B-C we include the experimental replicates as a new Supplementary Figure (see new Fig S5). Data in Figure S5 is now updated to include the experimental replicates. For the patient data in Figure 5, we have included details of specificity and sensitivity analysis for clinical samples (see new Fig 5C).

    **Minor comments:**

    "This method is relatively faster but may generate false positives due to non-specific amplification and primer interactions." Reference is needed.

    __Response: __We have now added the following references in support of this statement. (Gadkar, Goldfarb, Gantt, & Tilley, 2018; Sahoo, Sethy, Mohapatra, & Panda, 2016)

    "using the softwares Primer 3 and NUPACK." Reference is needed.

    __Response: __We have now added the following references (Untergasser et al., 2012; Zadeh et al., 2011)

    Reference 15 belongs to CRISPR-CAS based assay but it was cited under RT-LAMP assay.

    __Response: __This has now been corrected. We thank reviewer for this.

    Reviewer #1 (Significance (Required)):

    This paper will be of interest to scientists interested in developing diagnostic tools for the detection of SARS-CoV2 in viral and host pathogenic sequences; genetic disorders and development of precision medicine.

    Reviewer works in the field of Chemical Biology and Nanotechnology including sensor development and the application in diagnosis, cell physiological studies.

    Reviewer #3 (Evidence, reproducibility and clarity (Required)):

    In this manuscript, Charkravarthy et al. report a new method for detecting SARS-CoV-2 RNA in both in vitro and human saliva and nasal samples. The new detection method, PHANTOM, is capable of detecting as few as 100 copies of the SARS-CoV-2 genome. The method is demonstrated to reproducible over a large range of viral titers and results in a binary report on CoV-2 infection. From my perspective the results are strong and fairly convincing (please see comments below). There is clear, logical, flow to the experiments and engineering of the PHANTOM system. The collaborative work is well organized and logical. The work is clearly of high significance and certainly merits expedited review and publication. I would like to unambiguously state that support publication of this manuscript in its current form in the non-peer reviewed context of this journal, would be more than happy to provide further peer review of this manuscript upon submission to another journal, and would be more than happy to provide further comments if requested by the authors.

    My personal background is broad in range, however, I have a long track record of research in RNA folding, structural biology, biosensor development, and bioinformatics. Given this knowledge base, I found the manuscript rather easy to read and digest. The manuscript is well written and clear. In order to expedite the process of review I will not give a detailed review which would include grammatical errors (there are are very few). Rather, I will touch on the most pressing issues I see.

    **Major concerns:**

    1) There a number of figures that do not show a statistical measure of significance (e.g. error bears, ANOVA, etc.). It is essential that these be included in the final peer reviewed publication. (See Figure 2A, Figure 3D, Figure 4B, Figure 4C, Figure 5A, Figure 5C, Figure 5D).

    __Response: __This is an important point. We have added the error bars and statistical analyses to all relevant plots. We have included the description of these statistical parameters in the figure legends throughout the manuscript, where relevant. Alternately, experimental replicates are indicated.

    Specifically in Figures 2 and 3 and 4D we have performed statistical analysis to include p-values to show significance of the data. For the data in Figure 4 B-C we include the experimental replicates as a new Supplementary Figure (see new Fig S5). Data in Figure S5 is now updated to include the experimental replicates. For the data in Figure 5, we have included details of specificity and sensitivity analysis for clinical samples (see new Fig 5C).

    2) There are some important points that do not include references within the manuscript. I believe that the authors should reference Abdolahzadeh et al. RNA 2019 in the introduction. This manuscript describes another NASBA viral detection system using fluorescent RNA reporters (also see Trachman et al. Q. Rev. Biophys 2019, for reference on fluorescent aptamers). Also see the ROSALIND method (Jung et al. 2020 Nature Biotechnology) for detecting water contaminants using visual identification by fluorescent aptamers.

    __Response: __We have added the above mentioned references to the manuscript as suggested by the reviewer.

    3) The discussion states that "The overall sensitivity in the attomolar range ensures detection of infection in the majority of Covid-positive patients in a population". Please provide a reference to support this and explicitly state the concentration of viral RNA in patient samples. There are a number of times that the copy number of viral genomes and sensitivity of the measurement is stated throughout the manuscript. There should also be a reference and statement about concentration.

    __Response: __The reviewer has raised multiple connected points here, which we address in the revised manuscript.

    1. Concentration of RNA in patient samples: We have added the references (Pujadas et al., 2020; Wyllie et al., 2020) where the authors report that the typical concentration of viral RNA in patient nasopharyngeal swab samples lies in the range of 104 to 105 copies of RNA per ml. This translates to a concentration range of 10 to 100 attoMolar. This reference is now added to the manuscript. For the patient samples used on our study, we refer to the Ct- values obtained from the RT-PCR tests and correlate Ct values to the readout from our assay, consistent with other reports on COVID-19 diagnostics ((Joung et al., 2020; Vogels et al. 2020; Wu et al., 2021).

    Copy number and sensitivity: As the reviewer notes, we refer to viral genome copy number and sensitivity of our assay in the manuscript. These calculations of copy number and sensitivity were made using a commercial source of synthetic CoV-2 RNA (Twist Biosciences) that is used in several studies about COVID-19 diagnostics (Joung et al., 2020; Rabe & Cepko, 2020; Wu et al., 2021). The RNA copy numbers are taken from the product details provided by the manufacturer. These details are now clearly stated in the manuscript. The commercial RNA is provided at 106 copies per ul. From this, we take as low as 100 copies per 20ul of NASBA reaction, which we are able to detect using our assay. Hence our sensitivity comes to 8 attoMolar. We have clarified this in the manuscript. __We noticed a typo in the original submission where we refer to a sensitivity of 80 attomolar in the Discussion. This is corrected to 8 attomolar. __With this sensitivity we are within the range to detect RNA in patient samples, as confirmed by our patient data.

    Reviewer #3 (Significance (Required)):

    I think this is a significant advancement in the field. The introduction of smartphone technology to this robust diagnostic is very attractive. The work is of high significance since the researchers demonstrated robust responses against SARS-CoV-2 variants. As well all now know these are on the rise and cheap robust detection methods are essential for containing this virus.

    Response: We thank the reviewers for the positive comments.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    In this manuscript, Charkravarthy et al. report a new method for detecting SARS-CoV-2 RNA in both in vitro and human saliva and nasal samples. The new detection method, PHANTOM, is capable of detecting as few as 100 copies of the SARS-CoV-2 genome. The method is demonstrated to reproducible over a large range of viral titers and results in a binary report on CoV-2 infection. From my perspective the results are strong and fairly convincing (please see comments below). There is clear, logical, flow to the experiments and engineering of the PHANTOM system. The collaborative work is well organized and logical. The work is clearly of high significance and certainly merits expedited review and publication. I would like to unambiguously state that support publication of this manuscript in its current form in the non-peer reviewed context of this journal, would be more than happy to provide further peer review of this manuscript upon submission to another journal, and would be more than happy to provide further comments if requested by the authors.

    My personal background is broad in range, however, I have a long track record of research in RNA folding, structural biology, biosensor development, and bioinformatics. Given this knowledge base, I found the manuscript rather easy to read and digest. The manuscript is well written and clear. In order to expedite the process of review I will not give a detailed review which would include grammatical errors (there are are very few). Rather, I will touch on the most pressing issues I see.

    Major concerns:

    1. There a number of figures that do not show a statistical measure of significance (e.g. error bears, ANOVA, etc.). It is essential that these be included in the final peer reviewed publication. (See Figure 2A, Figure 3D, Figure 4B, Figure 4C, Figure 5A, Figure 5C, Figure 5D).

    2. There are some important points that do not include references within the manuscript. I believe that the authors should reference Abdolahzadeh et al. RNA 2019 in the introduction. This manuscript describes another NASBA viral detection system using fluorescent RNA reporters (also see Trachman et al. Q. Rev. Biophys 2019, for reference on fluorescent aptamers). Also see the ROSALIND method (Jung et al. 2020 Nature Biotechnology) for detecting water contaminants using visual identification by fluorescent aptamers.

    3. The discussion states that "The overall sensitivity in the attomolar range ensures detection of infection in the majority of Covid-positive patients in a population". Please provide a reference to support this and explicitly state the concentration of viral RNA in patient samples. There are number of times that the copy number of viral genomes and sensitivity of the measurement is stated throughout the manuscript. There should also be a reference and statement about concentration.

    Significance

    I think this is a significant advancement in the field. The introduction of smart phone technology to this robust diagnostic is very attractive. The work is of high significance since the researchers demonstrated robust reposes against SARS-CoV-2 variants. As well all now know these are on the rise and cheap robust detection methods are essential for containing this virus.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    In this project, authors develop a colorimetric and luminescence assay for the detection of SARS-CoV-2 RNA in vitro. They design an RNA based sensor that will be triggered by target RNA then release the ribosome binding site and a translation start site followed by a reporter gene. The released sequence will then trigger the production of reporter protein by transcription-translation coupled assay. Authors also introduce an RNA amplification step in order to increase the sensitivity of this assay.

    Strengths:

    This assay provides a simple, rapid way to detect SARS-CoV2 and it is an elegant way to incorporate transcription-translation coupled assay for SARS-CoV-2 RNA detection and identify SARS-CoV-2 patient samples. It is a nice assay and the performance is comparable with the existing method.

    Weaknesses:

    However, the positioning of this assay is not very clear. The readout of this assay could be recorded by camera whereas it includes several steps such as RNA extraction, amplification, transcription-translation coupled assay and reporter reaction. The limitations of the existing methods (RT-PCR, paper strip) and the advantages of this assay haven't been demonstrated by the experiments. The stability of RNA may also restrict the application of the proposed assay on site.

    Major comments:

    Authors are suggested to design an experiment to show the advantage of this assay compared with the existing method.

    What is the limit of detection of this assay using LacZ and Luciferase reporter respectively?

    Authors have not examined the selectivity of this assay. What is the specificity, selectivity for each of these variants? Do altering target RNA change the specificity?

    In Figure 2C-F, sensor 17 showed higher fold change and sensitivity. Why was sensor 12 selected for further study in Figure 3?

    Authors should show the error bar in all plots. Authors should also indicate what the error bar means (SD, S.E.M. etc.) throughout the manuscript.

    Minor comments:

    "This method is relatively faster but may generate false positives due to non-specific amplification and primer interactions." "using the softwares Primer 3 and NUPACK." Reference is needed.

    Reference 15 belongs to CRISPR-CAS based assay but it was cited under RT-LAMP assay.

    Significance

    This paper will be of interest to scientists interested in developing diagnostic tools for the detection of SARS-CoV2 in viral and host pathogenic sequences; genetic disorders and development of precision medicine.

    Reviewer works in the field of Chemical Biology and Nanotechnology including sensor development and the application in diagnosis, cell physiological studies.

  5. Version published to 10.1101/2021.01.08.21249426v3 on medRxiv
  6. ScreenIT

    SciScore for 10.1101/2021.01.08.21249426: (What is this?)

    Please note, not all rigor criteria are appropriate for all manuscripts.

    Table 1: Rigor

    NIH rigor criteria are not applicable to paper type.

    Table 2: Resources

    Software and Algorithms
    SentencesResources
    An Indian strain of SARS-CoV-2 (Accession Code MT012098.1) was downloaded from NCBI and used for analysis.
    NCBI
    suggested: (NCBI, RRID:SCR_006472)
    All primers were scored using Primer3 (v. 2.5.044), and NUPACK (v. 3.2.245).
    Primer3
    suggested: (Primer3, RRID:SCR_003139)
    Their trigger sequences were checked for potential similarity to the human genome (Accession code: GCF_000001405.38) and transcriptome (Refseq, 16May2020) using the megablast module of NCBIBLAST, with default parameters (e-value threshold: 0.05, gap costs: creation -5 extension -2, match/mismatch score: +2/-3).
    Refseq
    suggested: (RefSeq, RRID:SCR_003496)
    All data was plotted using GraphPad Prism 8 and figures were made using Adobe Illustrator.
    GraphPad Prism
    suggested: (GraphPad Prism, RRID:SCR_002798)
    Adobe Illustrator
    suggested: (Adobe Illustrator, RRID:SCR_010279)
    Image was further processed and analysed using Fiji/image J 1.52p.46 First, the RGB image was cropped into the required dimension.
    Fiji/image J
    suggested: None
    Amplicon and Trigger single strandedness were estimated.
    Amplicon
    suggested: (Amplicon, RRID:SCR_003294)

    Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

    Results from LimitationRecognizer: An explicit section about the limitations of the techniques employed in this study was not found. We encourage authors to address study limitations.

    Results from TrialIdentifier: No clinical trial numbers were referenced.

    Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

    Results from JetFighter: We did not find any issues relating to colormaps.

    Results from rtransparent:
    • Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
    • Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
    • No protocol registration statement was detected.

    About SciScore

    SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

  8. ScreenIT

    SciScore for 10.1101/2021.01.08.21249426: (What is this?)

    Please note, not all rigor criteria are appropriate for all manuscripts.

    Table 1: Rigor

    NIH rigor criteria are not applicable to paper type.

    Table 2: Resources

    Software and Algorithms
    SentencesResources
    An Indian strain of SARS-CoV-2 (Accession Code MT012098.1) was downloaded from NCBI and used for analysis.
    NCBI
    suggested: (NCBI, RRID:SCR_006472)
    All primers were scored using Primer3 (v. 2.5.044), and NUPACK (v. 3.2.245).
    Primer3
    suggested: (Primer3, RRID:SCR_003139)
    Their trigger sequences were checked for potential similarity to the human genome (Accession code: GCF_000001405.38) and transcriptome (Refseq, 16May2020) using the megablast module of NCBIBLAST, with default parameters (e-value threshold: 0.05, gap costs: creation -5 extension -2, match/mismatch score: +2/-3).
    Refseq
    suggested: (RefSeq, RRID:SCR_003496)
    All data was plotted using GraphPad Prism 8 and figures were made using Adobe Illustrator.
    GraphPad Prism
    suggested: (GraphPad Prism, RRID:SCR_002798)
    Adobe Illustrator
    suggested: (Adobe Illustrator, RRID:SCR_010279)
    Image was further processed and analysed using Fiji/image J 1.52p.46 First, the RGB image was cropped into the required dimension.
    Fiji/image J
    suggested: None
    Amplicon and Trigger single strandedness were estimated.
    Amplicon
    suggested: (Amplicon, RRID:SCR_003294)

    Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

    Results from LimitationRecognizer: An explicit section about the limitations of the techniques employed in this study was not found. We encourage authors to address study limitations.

    Results from TrialIdentifier: No clinical trial numbers were referenced.

    Results from Barzooka: We found bar graphs of continuous data. We recommend replacing bar graphs with more informative graphics, as many different datasets can lead to the same bar graph. The actual data may suggest different conclusions from the summary statistics. For more information, please see Weissgerber et al (2015).

    Results from JetFighter: We did not find any issues relating to colormaps.

    About SciScore

    SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

  9. Version published to 10.1101/2021.01.08.21249426v1 on medRxiv
  10. Version published to 10.1101/2021.01.08.21249426v2 on medRxiv