Dear Dr. Rehm,

Thank you very much for submitting your manuscript "Applying GAN-based data augmentation to improve transcriptome-based prognostication in breast cancer" for consideration at PLOS Computational Biology.

As with all papers reviewed by the journal, your manuscript was reviewed by members of the editorial board and by several independent reviewers. In light of the reviews (below this email), we would like to invite the resubmission of a significantly-revised version that takes into account the reviewers' comments.

We cannot make any decision about publication until we have seen the revised manuscript and your response to the reviewers' comments. Your revised manuscript is also likely to be sent to reviewers for further evaluation.

When you are ready to resubmit, please upload the following:

[1] A letter containing a detailed list of your responses to the review comments and a description of the changes you have made in the manuscript. Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

[2] Two versions of the revised manuscript: one with either highlights or tracked changes denoting where the text has been changed; the other a clean version (uploaded as the manuscript file).

Important additional instructions are given below your reviewer comments.

Please prepare and submit your revised manuscript within 60 days. If you anticipate any delay, please let us know the expected resubmission date by replying to this email. Please note that revised manuscripts received after the 60-day due date may require evaluation and peer review similar to newly submitted manuscripts.

Thank you again for your submission. We hope that our editorial process has been constructive so far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Zhaolei Zhang

Academic Editor

PLOS Computational Biology

Jian Ma

Section Editor

PLOS Computational Biology

***********************

Reviewer's Responses to Questions

Comments to the Authors:

Please note here if the review is uploaded as an attachment.

Reviewer #1: This is an interesting study in which the authors develop a predictive algorithm (T-GAN-D) that serves both as a classification and as a data augmentation method, with the goal of improving the stratification of low and high risk breast cancer patients from transcriptomic data. The proposed method is evaluated on two public gene expression datasets, namely the METABRIC and TCGA cohorts. Additionally, the predictive capabilities of the T-GAN-D are compared with the performance achieved by a CNN and several established breast cancer biomarkers. Finally, the work is reproducible, since all the material needed to replicate the results obtained in this study is publicly available to the scientific community.

However, some important concerns must be addressed to meet the publication criteria:

1. The rationale of converting gene expression data to arrays of numeric values needs to be further justified. In contrast with an image, a gene expression sample has an unstructured nature and lacks any local information that can be exploited by convolutional layers. In this way, authors should not only compare the performance of the T-GAN-D with a CNN, but also with the performance obtained by traditional ML models (such as SVM, random forest, etc.) that use gene expression vectors (without any image transformation) as input data. To counteract the great imbalance between the number of variables (high) and the number of samples (low), authors could use traditional feature selection methods to reduce the number of variables used as input to the classical ML classifiers.

2. Accuracy is used as a metric to evaluate the classification performance of T-GAN-D and CNN. However, additional metrics, e.g., specificity, sensitivity, should also be computed to perform a thorough evaluation of the classification capabilities of the models, by evaluating their performance separately for each of the two classes (high and low risk). This is specially relevant when evaluating the models on the TCGA dataset (last subsection of Results section), where there is a considerable difference between the number of high and low risk patients.

Minor comments:

- Line 243: “and” should be changed for “an”.

- Fig. 1F: “Predicted” should be removed from the legend of the figure.

- In Fig. 4A, authors need to clarify the meaning of “Pre” and “Post” transcripts filtering.

Reviewer #2: This manuscript proposed to use of a GAN-based data augmentation strategy to alleviate the overfitting of the classification model, with the aim of predicting the prognostic risk of breast cancer patients. The manuscript is in general well-written but needs to be checked thoroughly. I may have some concerns that need to be addressed.

Can you provide the number of patients and genes in both the MB cohort and TCGA cohort before and after defining the low and high-risk categories?

Please explain the rationality of the category definition of high-risk and low-risk patients.

How to convert the expression values of 18543 genes into the matrix of size 144*144?

What is the difference in time complexity between T-GAN-D and CNN?

It shows 161 patients are predicted low-risk group and high-risk group in both Fig.5 (B and C). I think the experiments resulted from the TCGA-BRCA dataset instead of MB + TCGA, or MB.

Current methods under comparison do not provide a conclusive benchmark.

Many descriptions are not standardized. such as inconsistent words (BRCA-TCGA and TCGA-BRCA).

Reviewer #3: The paper shows the application of a generative adversarial network (GAN) to improve prognosis in breast cancer using deep learning (DL). Specifically, the authors use transcriptomic data to identify low-risk and high-risk breast cancer patients. They process the gene expression data to turn it into images that can be more easily used by two DL models: a convolutional network (CNN) and an Auxiliary Classifier Wassertein GAN with gradient penalty (AC-WGAN-GP). The METABRIC and TCGA-BRCA selected cohorts are two of the largest and best-annotated breast cancer databases, an excellent choice for testing the efficacy of models. The paper shows that the discriminator of the AC-WGAN-GP model trained with the real processed images and the synthetic images created by the generator, as is logical in the normal process of a GAN, can be used as a DL classifier model to identify the patient prognosis. This discriminator, named by the authors as T-WGAN-D, improves the results obtained by a CNN. It should be noted that the paper does not present a new model or data augmentation (DA) application, but rather an application variant of part of an AC-WGAN model (the discriminator). The paper uses a non-precise definition of DA. It is not correct to consider DA the use of synthetic images in the training of a model that precisely requires these for its training. If the generated synthetic images were also used to augment the training set of another model, such as the CNN, it could be considered DA. Beyond this critique, the application of T-WGAN-D is novel enough considering the problem posed and it improves the performance achieved by the CNN model.

I pointed out some issues that needs to be addressed:

- A paragraph that explains structure of paper can be added at the end of the introduction section.

- Other references that also studied GAN-based data augmentation techniques on TCGA mRNA data can be added.

o Moreno-Barea, F.J., Jerez, J.M., Franco, L. (2022). GAN-Based Data Augmentation for Prediction Improvement Using Gene Expression Data in Cancer. In International Conference of Computational Science – ICCS 2022. Lecture Notes in Computer Science, vol 13352. Springer, Cham. doi: 10.1007/978-3-031-08757-8_3

o Kwon C, Park S, Ko S, Ahn J. Increasing prediction accuracy of pathogenic staging by sample augmentation with a GAN. PLoS One. 2021 Apr 27;16(4):e0250458. doi: 10.1371/journal.pone.0250458

- The authors should compare the performance of T-WGAN-D against other traditional methods and models used to identify high-risk breast cancer patients.

- The authors should try data augmentation on CNN, as increased variety may lead to improved performance. DA involves using synthetic data to train models that do not use it primarily.

- At line 243, the authors wrote " Introducing and independent cohort improves MB patient classification." Please review this error.

- In Figure 4.A it appears that AC-GAN was used, but T-GAN-D is mentioned in the text. Please review it.

- In the section on the results of the union of both cohorts to predict on one of them, the re-training of only the discriminator is mentioned. However, in Figures 3 and 5 the generator training also appears. These create confusion please revise accordingly.

- It would be convenient to mention transfer learning in the section described in the previous point. The retraining procedure is transfer learning, not DA.

- Future studies can be added with possibility of employing other deep learning architectures.

**********

Have the authors made all data and (if applicable) computational code underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data and code underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data and code should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data or code —e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

Figure Files:

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org.

Data Requirements:

Please note that, as a condition of publication, PLOS' data policy requires that you make available all data used to draw the conclusions outlined in your manuscript. Data must be deposited in an appropriate repository, included within the body of the manuscript, or uploaded as supporting information. This includes all numerical values that were used to generate graphs, histograms etc.. For an example in PLOS Biology see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5.

Reproducibility:

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Dear Dr. Rehm,

Thank you very much for submitting your manuscript "Applying a GAN-based classifier to improve transcriptome-based prognostication in breast cancer" for consideration at PLOS Computational Biology. As with all papers reviewed by the journal, your manuscript was reviewed by members of the editorial board and by several independent reviewers. The reviewers appreciated the attention to an important topic. Based on the reviews, we are likely to accept this manuscript for publication, providing that you modify the manuscript according to the review recommendations.

Please prepare and submit your revised manuscript within 30 days. If you anticipate any delay, please let us know the expected resubmission date by replying to this email.

When you are ready to resubmit, please upload the following:

[1] A letter containing a detailed list of your responses to all review comments, and a description of the changes you have made in the manuscript. Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out

[2] Two versions of the revised manuscript: one with either highlights or tracked changes denoting where the text has been changed; the other a clean version (uploaded as the manuscript file).

Important additional instructions are given below your reviewer comments.

Thank you again for your submission to our journal. We hope that our editorial process has been constructive so far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Zhaolei Zhang

Academic Editor

PLOS Computational Biology

Jian Ma

Section Editor

PLOS Computational Biology

***********************

A link appears below if there are any accompanying review attachments. If you believe any reviews to be missing, please contact ploscompbiol@plos.org immediately:

Reviewer's Responses to Questions

Comments to the Authors:

Please note here if the review is uploaded as an attachment.

Reviewer #1: Overall, the authors have produced an improved revised version of the manuscript. However, some concerns must be addressed to meet the publication criteria:

1. If we compare the results obtained by the traditional machine learning (ML) models (Supplementary Figures 5 and 6, as well as Supplementary Table 4) and the proposed T-GAN-D model, the T-GAN-D approach does not outperform the support vector machine (SVM) algorithm. Consequently, authors should justify in the manuscript the development of a complex deep learning (DL)-based methodology when a simpler and straightforward approach based on a feature selection method and a classical ML algorithm achieves an equivalent level of performance.

2. Authors should describe in the manuscript how sensitivity and specificity metrics were calculated, i.e. in the context of binary classification, which category (either high or low risk) was considered as positive, and which one was considered as negative.

Reviewer #2: The authors have answered my comments clearly.

Reviewer #3: The article shows many improvements compared to the first revised version. All the issues raised by the reviewers have been addressed and the scientific and technical quality of the article has increased with the changes introduced.

**********

Have the authors made all data and (if applicable) computational code underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data and code underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data and code should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data or code —e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

Figure Files:

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org.

Data Requirements:

Please note that, as a condition of publication, PLOS' data policy requires that you make available all data used to draw the conclusions outlined in your manuscript. Data must be deposited in an appropriate repository, included within the body of the manuscript, or uploaded as supporting information. This includes all numerical values that were used to generate graphs, histograms etc.. For an example in PLOS Biology see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5.

Reproducibility:

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

References:

Review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Dear Dr. Rehm,

We are pleased to inform you that your manuscript 'Applying a GAN-based classifier to improve transcriptome-based prognostication in breast cancer' has been provisionally accepted for publication in PLOS Computational Biology.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. A member of our team will be in touch with a set of requests.

Please note that your manuscript will not be scheduled for publication until you have made the required changes, so a swift response is appreciated.

IMPORTANT: The editorial review process is now complete. PLOS will only permit corrections to spelling, formatting or significant scientific errors from this point onwards. Requests for major changes, or any which affect the scientific understanding of your work, will cause delays to the publication date of your manuscript.

Should you, your institution's press office or the journal office choose to press release your paper, you will automatically be opted out of early publication. We ask that you notify us now if you or your institution is planning to press release the article. All press must be co-ordinated with PLOS.

Thank you again for supporting Open Access publishing; we are looking forward to publishing your work in PLOS Computational Biology. 

Best regards,

Zhaolei Zhang

Academic Editor

PLOS Computational Biology

Jian Ma

Section Editor

PLOS Computational Biology

***********************************************************