The proto-oncogene MYC is frequently amplified in human cancers and encodes the transcription factor c-Myc, which is associated with a variety of cellular processes such as cell growth and proliferation (Dang, 2013). While overexpressed c-Myc is known to contribute to tumorigenesis, an understanding of how this process occurs is complicated by a number of issues, including the large number of binding sites and the diversity between systems. While this was thought to occur by regulation of a specific subset of genes, Lin et al. (2012) present findings that c-Myc functions by globally amplifying the expression of actively transcribed genes.

The system used is the human P493-6 B cell model of Burkitt's lymphoma, which contains a tetracycline-repressible MYC transgene, allowing for titration of c-Myc protein (Schuhmacher et al., 1999; Pajic et al., 2000). The levels of c-Myc can be reduced and subsequently re-induced in a gradual time-dependent manner, as determined by western blot, which is shown in Figure 1B (Lin et al., 2012). As soon as 1 hr after re-induction, the protein levels of c-Myc increased above the repressed levels, which by 24 hr were similar to the tetracycline-free condition (Lin et al., 2012). This system has been used in other studies with similar results observed (Schuhmacher et al., 1999; Pajic et al., 2000). This experiment is important to replicate because it assesses the level of c-Myc re-induction with this system that will be used in the following experiments. This experiment is replicated in Protocol 1.

In Figure 3E the total levels of RNA, in P493-6 cells before and after re-induction of c-Myc, was determined by UV/VIS spectrophotometry (Lin et al., 2012). Lin et al. (2012) reported an increase in levels of absolute RNA over the timecourse of c-Myc re-induction. This experiment shows that c-Myc increases RNA content per cell, indicating c-Myc functions primarily in transcriptional amplification. Related experiments, using mouse B cells, also observed the same c-Myc-dependent amplification of cellular RNA content (Nie et al., 2012; Sabò et al., 2014). However, similar experiments in 3T9 fibroblasts or U2OS cells expressing an inducible c-Myc did not observe an increase in cellular RNA content (Sabò et al., 2014; Walz et al., 2014). Interestingly, an increase in total cellular RNA content was observed in 3T9 fibroblasts following serum stimulation, which was not observed when c-Myc was deleted in these cells (Sabò et al., 2014). This experiment is replicated in Protocol 2.

In Figure 3F, Lin et al. (2012) examined the transcriptional profile of a large number of genes from multiple functional categories in cells before and after re-induction of c-Myc using digital gene expression. Re-induction of c-Myc increased the number of active genes (defined as greater than 1 transcript/cell), but did not alter silent genes (defined as less than 0.5 transcript/cell) (Lin et al., 2012). This key finding suggests that elevated c-Myc levels lead to an amplification of the existing transcriptional profile. This finding was observed in other experiments using mouse B cells treated with the Myc-Max dimerization inhibitor 10058-F4 and analyzed by ChIP-Seq (Nie et al., 2012). Another report using P493-6 cells also found an increase in transcription of c-Myc target genes when c-Myc re-induction was titrated to different concentrations by microarray analysis (Schuhmacher and Eick, 2013). This experiment is replicated in Protocols 3 and 4. Recently, two papers were published that used primarily RNA-seq and ChIP-seq to focus on assessing if the transcriptional effects of c-Myc are direct or indirect (Alderton, 2014). These studies found that RNA amplification and promoter/enhancer invasion by c-Myc were separable events in 3T9 fibroblasts and U2OS cells, suggesting that c-Myc regulates a distinct subset of genes, which indirectly lead to RNA amplification (Sabò et al., 2014; Walz et al., 2014).

1. 1. Alderton GK

   (2014) Transcription: the transcriptional effects of MYC

   Nature Reviews Cancer 14:513.

   https://doi.org/10.1038/nrc3790

   • Google Scholar
2. 1. Dang CV

   (2013) MYC, metabolism, cell growth, and tumorigenesis

   Cold Spring Harbor Perspectives in Medicine 3:a014217.

   https://doi.org/10.1101/cshperspect.a014217

   • Google Scholar
3. 1. Faul F
   2. Erdfelder E
   3. Lang AG
   4. Buchner A

   (2007) G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences

   Behavior Research Methods 39:175–191.

   https://doi.org/10.3758/BF03193146

   • Google Scholar
4. 1. Lakens D

   (2013) Calculating and reporting effect sizes to facilitate cumulative science: a practical primer for t-tests and ANOVAs

   Frontiers in Psychology 4:863.

   https://doi.org/10.3389/fpsyg.2013.00863

   • Google Scholar
5. 1. Lin CY
   2. Lovén J
   3. Rahl PB
   4. Paranal RM
   5. Burge CB
   6. Bradner JE
   7. Lee TI
   8. Young RA

   (2012) Transcriptional amplification in tumor cells with elevated c-Myc

   Cell 151:56–67.

   https://doi.org/10.1016/j.cell.2012.08.026

   • Google Scholar
6. 1. Nie Z
   2. Hu G
   3. Wei G
   4. Cui K
   5. Yamane A
   6. Resch W
   7. Wang R
   8. Green DR
   9. Tessarollo L
   10. Casellas R
   11. Zhao K
   12. Levens D

   (2012) c-Myc is a universal amplifier of expressed genes in lymphocytes and embryonic stem cells

   Cell 151:68–79.

   https://doi.org/10.1016/j.cell.2012.08.033

   • Google Scholar
7. 1. Pajic A
   2. Spitkovsky D
   3. Christoph B
   4. Kempkes B
   5. Schuhmacher M
   6. Staege MS
   7. Brielmeier M
   8. Ellwart J
   9. Kohlhuber F
   10. Bornkamm GW
   11. Polack A
   12. Eick D

   (2000) Cell cycle activation by c-myc in a burkitt lymphoma model cell line

   International Journal of cancer. Journal International Du Cancer 87:787–793.

   https://doi.org/10.1002/1097-0215(20000915)87:63.0.CO;2-6

   • Google Scholar
8. 1. Sabò A
   2. Kress TR
   3. Pelizzola M
   4. de Pretis S
   5. Gorski MM
   6. Tesi A
   7. Morelli MJ
   8. Bora P
   9. Doni M
   10. Verrecchia A
   11. Tonelli C
   12. Fagà G
   13. Bianchi V
   14. Ronchi A
   15. Low D
   16. Müller H
   17. Guccione E
   18. Campaner S
   19. Amati B

   (2014) Selective transcriptional regulation by Myc in cellular growth control and lymphomagenesis

   Nature 511:488–492.

   https://doi.org/10.1038/nature13537

   • Google Scholar
9. 1. Schlosser I
   2. Hölzel M
   3. Hoffmann R
   4. Burtschler H
   5. Kohlhuber F
   6. Schuhmacher M
   7. Chapman R
   8. Weidle UH
   9. Eick D

   (2005) Dissection of transcriptional programmes in response to serum and c-Myc in a human B-cell line

   Oncogene 24:520–524.

   https://doi.org/10.1038/sj.onc.1208198

   • Google Scholar
10. 1. Schuhmacher M
    2. Staege MS
    3. Pajic A
    4. Polack A
    5. Weidle UH
    6. Bornkamm GW
    7. Eick D
    8. Kohlhuber F

    (1999) Control of cell growth by c-Myc in the absence of cell division

    Current Biology 9:1255–1258.

    https://doi.org/10.1016/S0960-9822(99)80507-7

    • Google Scholar
11. 1. Schuhmacher M
    2. Eick D

    (2013) Dose-dependent regulation of target gene expression and cell proliferation by c-Myc levels

    Transcription 4:192–197.

    https://doi.org/10.4161/trns.25907

    • Google Scholar
12. 1. Walz S
    2. Lorenzin F
    3. Morton J
    4. Wiese KE
    5. von Eyss B
    6. Herold S
    7. Rycak L
    8. Dumay-Odelot H
    9. Karim S
    10. Bartkuhn M
    11. Roels F
    12. Wüstefeld T
    13. Fischer M
    14. Teichmann M
    15. Zender L
    16. Wei CL
    17. Sansom O
    18. Wolf E
    19. Eilers M

    (2014) Activation and repression by oncogenic MYC shape tumour-specific gene expression profiles

    Nature 511:483–487.

    https://doi.org/10.1038/nature13473

    • Google Scholar

Thank you for sending your work entitled “Registered report: Transcriptional Amplification in Tumor Cells with Elevated c-Myc” for consideration at eLife. Your article has been evaluated by Sean Morrison (Senior editor), a Reviewing editor, and 3 reviewers, one of whom is a biostatistician.

The Reviewing editor and the reviewers discussed their comments before we reached this decision, and the Reviewing editor has assembled the following comments to help you prepare a revised submission.

As detailed below, the reviewers raised a number of major concerns that need to be addressed in a revised Registered report.

Major comments:

1) Any replication of the Lin et al. paper needs to include both RNA profiling and ChIP-seq experiments. The Lin et al. paper has been quite controversial and was based in very large part on ChIP-seq data and its interpretation. Many of the conclusions were based on very subtle changes in data profiles. Subsequently, two papers were published in Nature (Walz et al. and Sabo et al.) that challenge the global claims of Lin et al. In both papers, the authors worked hard to accumulate comprehensive ChIP-seq and RNA expression data in carefully designed experimental systems. While some aspects of the Lin et al. paper may be correct, both Nature papers conclude that there is a set of defined target genes that are far more Myc responsive than others. Hence, reproducing only a subset of the Lin et al. experiments is unlikely to add anything new or resolve controversial claims.

The authors do not propose to reproduce the critical ChIP-seq data and they do not propose any analysis of RNAPII profiles that would support or conflict with the conclusions of the Lin et al. paper, namely that Myc promotes genome-wide transcriptional elongation. The proposal only focuses on RNA profiling without integration with binding of Myc and RNAPII.

2) More comprehensive RNA-seq analysis would determine global RNA expression in response to Myc and not be limited to a subset of genes represented by NanoString.

3) The cell line P493-6 has been established 15 years ago. The proliferation of these cells depends on c-Myc expression and the presence of serum. Serum is a major variable in this system and the majority of genes in stimulated cells are regulated by serum and not by c-Myc (Schlosser et al., Oncogene, 2005). The impact of different serum changes on P493-6 cells is highly significant. In some serum batches the cells barely grow after c-Myc activation. Unfortunately, the impact of various serum batches on P493-6 cells has never been systematically analyzed. Moreover, meanwhile many batches of P493-6 cells are distributed worldwide. These cells have been cultured with different types of serum in different laboratories. Exposure to different sera probably has altered the epigenetic state of P493-6 cells further contributing to variation in gene expression.

From the scientific view, it would be more helpful to study the stability of this biological system, e.g. by culturing P493-6 cells over longer periods of time in different batches of serum followed by a subsequent transcriptome analysis +/- Myc. At minimum, the authors should perform their experiments using multiple batches of serum to assess whether this significantly alters their results.

Statistical comments to the authors:

4) For protocol 1 and 2, authors propose to use ANOVA to analyze the data. Please make sure that the data do not violate the assumptions of the ANOVA: normality and homoscedasiticity. If the data do not fit the assumptions well enough, try to find a data transformation that makes them fit. If this doesn't work, you will need to apply a nonparametric counterpart of ANOVA such as Kruskal-Wallis test. In addition, performing contrast within the framework of ANOVA is more powerful than performing a separate t-test if the assumption of ANOVA is valid.

5) Authors used G*Power to calculate the power. I think that power calculation for protocol 3 & 4 is probably based on the test family t-test implemented in G*Power since there is no Wilcoxon sum rank test implemented in the G*Power. I suggest using t-tests as test family and matched pairs as statistical test to recalculate the power for protocol 3 and 4 (see below for justification). You will need to re-compute the effect size by calculating the SD for paired design, although mean difference between two groups will stay the same regardless.

6) Authors propose to use two-tailed Wilcoxon sum rank test, which has been used in the original paper. I suggest use either two-tailed Wilcoxon signed rank test or two-tailed paired t-test. If you prefer use G* Power to calculate power, then you will be left with two-tailed paired t-test option. The reason why paired analysis is needed is that expressions of the same gene across different conditions are not independent.

7) One major conclusion from the original paper (Lin et al., 2012) is that elevated c-Myc in tumor cells leads to amplification of the expression of actively transcribed genes, but has no effect on silent genes. I am wondering whether the authors will perform the same test to the silent genes, as well as the actively transcribed genes, to confirm the results from the original paper.

8) While it is very useful to leverage the previously reported effects to compute minimum power a priori, what you really need is to guarantee a minimum power on your own data. This can be done, a priori, by including some cross-study variation. This will be helpful for you to plan on the number of replicates and so forth. Papers by Giovanni Parmigiani and collaborators at the Dana–Farber provide some estimates about cross-study variation that could be used for this purpose. Worst case, you should budget some additional variability because of cross-study reproducibility, and increase the sample size as appropriate. We also want you to compute and report power post-hoc/on-the-fly on your own data. Some minimum power should be guaranteed using summaries of your own data.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled “Registered report: Transcriptional Amplification in Tumor Cells with Elevated c-Myc” for further consideration at eLife. Your revised article has been favorably evaluated by Sean Morrison (Senior editor), a Reviewing editor, and the original reviewers. As you might expect, there was a mixed response from the reviewers regarding the changes. On balance, we would like to move forward but would ask you to make one additional change. Different serum batches have only been included for the c-Myc off situation. To complete this control, please also include different serum batches for the c-Myc on situation.

1) Any replication of the Lin et al. paper needs to include both RNA profiling and ChIP-seq experiments. The Lin et al. paper has been quite controversial and was based in very large part on ChIP-seq data and its interpretation. Many of the conclusions were based on very subtle changes in data profiles. Subsequently, two papers were published in Nature (Walz et al. and Sabo et al.) that challenge the global claims of Lin et al. In both papers, the authors worked hard to accumulate comprehensive ChIP-seq and RNA expression data in carefully designed experimental systems. While some aspects of the Lin et al. paper may be correct, both Nature papers conclude that there is a set of defined target genes that are far more Myc responsive than others. Hence, reproducing only a subset of the Lin et al. experiments is unlikely to add anything new or resolve controversial claims.

The authors do not propose to reproduce the critical ChIP-seq data and they do not propose any analysis of RNAPII profiles that would support or conflict with the conclusions of the Lin et al. paper, namely that Myc promotes genome-wide transcriptional elongation. The proposal only focuses on RNA profiling without integration with binding of Myc and RNAPII.

We agree that all of the experiments included in the original study are important, and choosing which experiments to replicate has been one of the great challenges of this project. The Reproducibility Project: Cancer Biology (RP:CB) aims to replicate experiments that are impactful, but does not necessarily aim to replicate all the impactful experiments in any given paper. In this case, the RP:CB core team felt that the RNA profiling experiment in Lin et al., 2012, was critical as it was a key part of the reported finding that Myc induction increased the expression of active genes but not silent genes, indicating the predominant effect of substantially elevated levels of Myc is amplified transcription of the existing gene expression program. We also agree the ChIP-seq experiments are a critical component of the overall reported finding, however these types of experiments (sequencing) are excluded from all articles. As this was only a part of the paper, the other experiments were still eligible. These exclusion criteria are outlined on the project page (https://osf.io/e81xl/wiki/studies) and in a Feature article describing the project (recently accepted for publication by eLife). We agree that the exclusion of certain experiments limits the scope of what can be analyzed by the project, but we are attempting to identify a balance of breadth of sampling for general inference with sensible investment of resources on replication projects to determine to what extent the included experiments are reproducible. Therefore, we will restrict our analysis to the experiments being replicated and will not include discussion of experiments not being replicated in this study.

We have also updated the Introduction to include the two papers (Walz et al. and Sabo et al.) that also examine transcriptional regulation by Myc.

2) More comprehensive RNA-seq analysis would determine global RNA expression in response to Myc and not be limited to a subset of genes represented by NanoString.

We agree RNA-seq analysis would be an informative approach to determine the global RNA expression in response to Myc, however Lin and colleagues did not use this approach. The Reproducibility Project: Cancer Biology aims to perform direct replications using the same methodology reported in the original paper. The use of RNA-seq analysis would be a conceptual replication, which we agree is a useful approach to test the experiment’s underlying hypothesis, but which is not an aim of the project.

3) The cell line P493-6 has been established 15 years ago. The proliferation of these cells depends on c-Myc expression and the presence of serum. Serum is a major variable in this system and the majority of genes in stimulated cells are regulated by serum and not by c-Myc (Schlosser et al., Oncogene, 2005). The impact of different serum changes on P493-6 cells is highly significant. In some serum batches the cells barely grow after c-Myc activation. Unfortunately, the impact of various serum batches on P493-6 cells has never been systematically analyzed. Moreover, meanwhile many batches of P493-6 cells are distributed worldwide. These cells have been cultured with different types of serum in different laboratories. Exposure to different sera probably has altered the epigenetic state of P493-6 cells further contributing to variation in gene expression.

From the scientific view, it would be more helpful to study the stability of this biological system, e.g. by culturing P493-6 cells over longer periods of time in different batches of serum followed by a subsequent transcriptome analysis +/- Myc. At minimum, the authors should perform their experiments using multiple batches of serum to assess whether this significantly alters their results.

This project focuses on direct replication of the experiments as detailed in the original report and with information provided by the original authors. Aspects of an experiment not included in the original study are occasionally added to ensure the quality of the research, but by no means is a requirement of this project; rather, it is an extension of the original work. Adding additional aspects not included in the original study can be of scientific interest, and can be included, if it is possible, to balance them with the main aim of this project: to perform a direct replication of the original experiment(s).

Therefore, we agree with the reviewers that there is scientific interest in better understanding the stability of this biological system. We will be using the same P493-6 cells used by the original authors (generously provided by Charles Lin) and will use the same source of Tet System Approved FBS as originally reported. However, even if we were to know the lot number of FBS originally used, it is unlikely we would be able to obtain it. We will use the same lot for all the experiments described, but have included an additional cohort to all experiments that will use a different lot of FBS. This cohort will be harvested in the same manner described for the cohort harvested 0 hr after tetracycline induction. These two cohorts, both 0 hr after tetracycline induction, but with different lots of FBS, will be compared to each other to assess if there is variability between different batches of serum.

Statistical comments to the authors:

4) For protocol 1 and 2, authors propose to use ANOVA to analyze the data. Please make sure that the data do not violate the assumptions of the ANOVA: normality and homoscedasiticity. If the data do not fit the assumptions well enough, try to find a data transformation that makes them fit. If this doesn't work, you will need to apply a nonparametric counterpart of ANOVA such as Kruskal-Wallis test. In addition, performing contrast within the framework of ANOVA is more powerful than performing a separate t-test if the assumption of ANOVA is valid.

Thank you for this suggestion. At the time of analysis, we will perform the Shapiro-Wilk test and generate a quantile-quantile (q-q) plot, to assess the normality of the data, and also perform the Brown-Forsythe test to assess homoscedasiticity. If the data appears skewed, we will perform the appropriate transformation in order to proceed with the proposed statistical analysis. We will note any changes or transformations made. If this doesn’t work we will perform the Kruskal-Wallis test and if necessary the Wilcoxon-Mann-Whitney test. We have also updated the manuscript to address this point.

Also, in protocol 1 we made the intended planned comparison (contrast) explicit to clarify that it is not a separate t-test.

5) Authors used G*Power to calculate the power. I think that power calculation for protocol 3 & 4 is probably based on the test family t-test implemented in G*Power since there is no Wilcoxon sum rank test implemented in the G*Power. I suggest using t-tests as test family and matched pairs as statistical test to recalculate the power for protocol 3 and 4 (see below for justification). You will need to re-compute the effect size by calculating the SD for paired design, although mean difference between two groups will stay the same regardless.

G*Power does have the Wilcoxon sum rank test as an option. It is called Wilcoxon-Mann-Whitney test, which is another name for this test. However, we agree with the recalculation for protocols 3 and 4 suggested below and now include the two-tailed Wilcoxon signed rank test instead of the Wilcoxon sum rank test.

6) Authors propose to use two-tailed Wilcoxon sum rank test, which has been used in the original paper. I suggest use either two-tailed Wilcoxon signed rank test or two-tailed paired t-test. If you prefer use G* power to calculate power, then you will be left with two-tailed paired t-test option. The reason why paired analysis is needed is that expressions of the same gene across different conditions are not independent.

We agree with this assessment. A two-tailed Wilcoxon signed rank test is the appropriate test because the data is paired and the normality assumption does not hold. We have recalculated power and sample size for protocols 3 and 4 accordingly. We also changed the language in the analysis plan and added this as a known difference from the original study.

7) One major conclusion from the original paper (Lin et al., 2012) is that elevated c-Myc in tumor cells leads to amplification of the expression of actively transcribed genes, but has no effect on silent genes. I am wondering whether the authors will perform the same test to the silent genes, as well as the actively transcribed genes, to confirm the results from the original paper.

We agree and have included the analysis of silent genes in the analysis plan and performed sensitivity calculations to determine the effect size that will be detected with 80% power. In the original paper (Lin et al., 2012), the authors determined actively transcribed genes as > 1 transcript/cell and silent genes as < 0.5 transcript/cell in cells with lower levels of c-Myc (0 hr after tet induction). However, this excludes 69 genes from the analysis that fall between 1 and 0.5. Thus, in the analysis plan we have included silent genes as defined by the original authors (< 0.5 transcript/cell) and also non-active genes (< 1 transcript/cell).

8) While it is very useful to leverage the previously reported effects to compute minimum power a priori, what you really need is to guarantee a minimum power on your own data. This can be done, a priori, by including some cross-study variation. This will be helpful for you to plan on the number of replicates and so forth. Papers by Giovanni Parmigiani and collaborators at the Dana–Farber provide some estimates about cross-study variation that could be used for this purpose. Worst case, you should budget some additional variability because of cross-study reproducibility, and increase the sample size as appropriate. We also want you to compute and report power post-hoc/on-the-fly on your own data. Some minimum power should be guaranteed using summaries of your own data.

We thank the reviewers for these suggestions. The cross-study variation, such as approaches that utilize the 95% confidence interval of the effect size, can be useful in conducting power calculations when planning adequate sample sizes for detecting the true population effect size, which requires a range of possible observed effect sizes. However, the Reproducibility Project: Cancer Biology is designed to conduct replications that have 80% power to detect the point estimate of the originally reported effect size. While this has the limitation of being underpowered to detect smaller effects than what is originally reported, this standardizes the approach across all studies to be designed to detect the originally reported effect size with at least 80% power. Also, while the minimum power guarantee is beneficial for observing a range of possible effect sizes, the experiments in this replication, and all experiments in the project, are designed to detect the originally reported effect size with a minimum power of 80%. Thus, performing power calculations during or after data collection is not necessary in this replication attempt as all studies included are already designed to meet a minimum power or are identified beforehand as being underpowered and thus are not included in the confirmatory analysis plan. The papers by Giovanni Parmigiani and collaborators highlight the importance of accounting for variability that can occur across different studies, specifically gene expression data. While it is possible for a difference in variance between the originally reported results and the replication data, this will be reflected in the presentation of the data and a possible reason for obtaining a different effect size estimate.

[Editors’ note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled ”Registered report: Transcriptional Amplification in Tumor Cells with Elevated c-Myc” for further consideration at eLife. Your revised article has been favorably evaluated by Sean Morrison (Senior editor), a Reviewing editor, and the original reviewers. As you might expect, there was a mixed response from the reviewers regarding the changes. On balance, we would like to move forward but would ask you to make one additional change. Different serum batches have only been included for the c-Myc off situation. To complete this control, please also include different serum batches for the c-Myc on situation.

We agree and have adjusted the manuscript. We will use two different lots of serum to grow the cells for all the experiments.

Author details

1. David Blum

   Bioexpression and Fermentation Facility, University of Georgia, Athens, Georgia

   Contribution

   DB, Drafting or revising the article

   Competing interests

   DB: This is a Science Exchange Associated lab.

2. Haiping Hao

   JHMI Deep Sequencing and Microarray Core Facility, Johns Hopkins University, Baltimore

   Contribution

   HH, Drafting or revising the article

   Competing interests

   HH: This is a Science Exchange Associated lab.

3. Michael McCarthy

   University of Oxford, Oxford, United Kingdom

   Contribution

   MMC, Drafting or revising the article

   Competing interests

   No competing interests declared.

4. Reproducibility Project: Cancer Biology

   Contribution

   RP:CB, Conception and design, Drafting or revising the article

   For correspondence

   tim@cos.io

   Competing interests

   RP:CB: EI, FT and JL are employed by and hold shares in Science Exchange Inc.

   1. Elizabeth Iorns, Science Exchange, Palo Alto, California
   2. William Gunn, Mendeley, London, United Kingdom
   3. Fraser Tan, Science Exchange, Palo Alto, California
   4. Joelle Lomax, Science Exchange, Palo Alto, California
   5. Timothy Errington, Center for Open Science, Charlottesville, Virginia

• 2,191

  Page views

• 129

  Downloads

• 6

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.