A graph neural network-based approach for predicting SARS-CoV-2–human protein interactions from multiview data

Sumanta Ray; Syed Alberuni; Alexander Schönhuth

doi:10.1371/journal.pone.0332794

Peer Review History

Original SubmissionApril 18, 2025
19 May 2025 Decision Letter - Chandrabose Selvaraj, Editor PONE-D-25-21081A Graph Learning Framework for Comprehensive Prediction of SARS-CoV-2 and Human Protein Interactions from Multiview Protein Interaction DataPLOS ONE Dear Dr. Ray, Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. Please submit your revised manuscript by Jul 03 2025 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. Please include the following items when submitting your revised manuscript: A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols. We look forward to receiving your revised manuscript. Kind regards, Chandrabose Selvaraj, Ph.D. Academic Editor PLOS ONE Journal requirements: When submitting your revision, we need you to address these additional requirements. 1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf 2. Please note that PLOS ONE has specific guidelines on code sharing for submissions in which author-generated code underpins the findings in the manuscript. In these cases, we expect all author-generated code to be made available without restrictions upon publication of the work. Please review our guidelines at https://journals.plos.org/plosone/s/materials-and-software-sharing#loc-sharing-code and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse. 3. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information. Additional Editor Comments: In case of reviewers recommending the citations that are not directly pertinent to the scope or content of the manuscript, authors are encouraged to provide a reasoned justification for declining such suggestions. The editorial board affirms that the exclusion of non-essential references will not influence the editorial decision regarding the manuscript. [Note: HTML markup is below. Please do not edit.] Reviewers' comments: Reviewer's Responses to Questions Comments to the Author 1. Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented. Reviewer #1: Yes Reviewer #2: Yes ******** 2. Has the statistical analysis been performed appropriately and rigorously? Reviewer #1: Yes Reviewer #2: Yes ****** 3. Have the authors made all data underlying the findings in their manuscript fully available? The PLOS Data policy requires authors to make all data 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 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—e.g. participant privacy or use of data from a third party—those must be specified. Reviewer #1: Yes Reviewer #2: Yes ****** 4. Is the manuscript presented in an intelligible fashion and written in standard English? PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here. Reviewer #1: No Reviewer #2: Yes ****** 5. Review Comments to the Author Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters) Reviewer #1: I have now completed the review of the manuscript titled “A Graph Learning Framework for Comprehensive Prediction of SARS-CoV-2 and Human Protein Interactions from Multiview Protein Interaction Data”, My specific comments are 1. The title is descriptive, but consider specifying the method (e.g., graph neural network-based) to clarify novelty. 2. The abstract needs clearer articulation of the main results and comparative performance metrics. Include key numerical outcomes (e.g., accuracy, precision) to support claims. 3. The introduction provides good background but lacks critical discussion of prior computational methods for SARS-CoV-2–human protein interactions. Consider referencing recent work such as https://doi.org/10.1371/journal.pone.0294769 ; https://doi.org/10.3389/fphar.2024.1369659 ; Virtual Screening and Molecular Docking of FDA Approved Antiviral Drugs for the Identification of Potential Inhibitors of SARS-CoV-2 RNA-MTase Protein; https://doi.org/10.1002/cbdv.202402548 4. Motivation for using multiview data and graph learning is justified but would benefit from a brief example of why single-view models may miss relevant interactions. 5. The integration of STRING, BioGRID, and sequence-based interactions is appreciated. However, the criteria for interaction confidence thresholding are not clearly explained. Were low-confidence interactions filtered? The authors can potentially benefir from https://doi.org/10.1021/acsomega.3c07866; https://doi.org/10.3389/fphar.2025.1509263 ; https://doi.org/10.3389/fphar.2025.1509263 ; https://doi.org/10.1007/s10989-020-10076-w 6. The process for fusing multiview networks via Wasserstein distance is conceptually interesting, but the explanation is too terse. Please provide a schematic or algorithmic pseudocode to support reproducibility. 7. The model is claimed to use GCN and Wasserstein integration, but details on the number of layers, training epochs, loss function, and hyperparameter tuning are missing. 8. It is unclear how negative samples were generated or balanced. Were negative protein pairs randomly sampled or drawn from known non-interacting pairs? 9. Figure 2 shows model architecture but lacks annotations and descriptions. It is hard to interpret the role of each module in the workflow. 10. There is no table presenting the model’s quantitative performance (e.g., accuracy, F1-score, AUC) on benchmark datasets. This weakens the credibility of performance claims. 11. Add comparative results against baseline methods (e.g., logistic regression, DeepWalk, GCN without multiview fusion). 12. The identification of known host proteins (e.g., ACE2, TMPRSS2) adds confidence, but the analysis is anecdotal. Perform GO enrichment or pathway analysis on the predicted top 100 proteins to support biological relevance. 13. Drug repurposing suggestions (e.g., dexamethasone, baricitinib) are mentioned, but how were these inferred from the predicted PPIs? The method of linking predictions to drugs should be more systematically described. 14. The conclusion is too generic. Reflect more critically on the limitations, such as the dependence on existing PPI databases and lack of experimental validation. 15. Future directions could include validation using CRISPR screens or cross-species PPI transfer learning. 16. The manuscript needs a thorough language edit for grammar and clarity. Examples: o Page 5: “This approach is very efficiently to...” → “This approach is efficient for...” o Page 8: “The ROC curve indicate the...” → “The ROC curve indicates. Reviewer #2: 1. Clustering for proteins can be overlapping in nature in biological data, but the authors use the hierarchical clustering which detects disjoint clusters and has a high time computational cost and also the method is old and could be error prone. – Proper justification require. 2. Algorithm also has high overall time complexity so it would be slow for large datasets. –Need more statement about it. 3. What is significance of you own algorithm in relation to other already available similar algorithms? A comparative analysis could be better for the paper and could enhance the novelty of the work. 4. These proteins already serve the purposes of host-directed therapy (HDT) options for other viruses, thereby also being potential candidates for building HDT strategy against SARS-CoV2. --- “thereby also being potential candidates for building HDT”—More explanation require. 5. In this connection, the list of drugs which we suggest to target the predicted proteins may hold strong promise for yielding repurposable drugs to use against COVID-19.---- more strong justification require. 6. Authors should check drug- protein interaction and calculate the free energy to make such conclusion. ****** 6. 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 ******** [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment 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. Registration is free. 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 PLOS at figures@plos.org. Please note that Supporting Information files do not need this step. https://doi.org/10.1371/journal.pone.0332794.r001
Revision 1
16 Aug 2025 Author Response Answer to comments of Reviewer 1 Comments: 1. The title is descriptive, but consider specifying the method (e.g., graph neural network-based) to clarify novelty. Answer: Thank you for the valuable suggestion. We have now revised the title to explicitly highlight the method as follows: "A Graph Neural Network-Based Approach for Predicting SARS-CoV-2–Human Protein Interactions from Multiview Data." 2. The abstract needs clearer articulation of the main results and comparative performance metrics. Include key numerical outcomes (e.g., accuracy, precision) to support claims. Answer: We appreciate this comment. The abstract now explicitly mentions our main results, including quantitative performance metrics. Specifically, our GCN model achieved test ROC-AUCs of 86.00 (PPI), 83.00 (GO similarity), and 83.10 (sequence similarity) and average precision (AP) scores of 86.10, 82.12, and 82.30, respectively (see Table 1). In total, we predicted 472 high-confidence interactions between 280 host and 27 viral proteins, and our drug-target analysis found strong links to repurposable drugs (see Table 5). These key numerical results are now stated in the abstract of the revised version of the manuscript. Please see the abstract of the revised version of the manuscript. 3. The introduction provides good background but lacks critical discussion of prior computational methods for SARS-CoV-2–human protein interactions. Consider referencing recent work such as https://doi.org/10.1371/journal.pone.0294769 ; https://doi.org/10.3389/fphar.2024.1369659 ; Virtual Screening and Molecular Docking of FDA Approved Antiviral Drugs for the Identification of Potential Inhibitors of SARS-CoV-2 RNA-MTase Protein; https://doi.org/10.1002/cbdv.202402548 Answer: Thank you for highlighting this gap. We have expanded the introduction to critically discuss prior computational methods for SARS-CoV-2–human protein interaction prediction, including recent approaches that integrate virtual screening, molecular docking, and sequence-based predictors ([4, 5, 6, references in the manuscript]). The following recent works have also been cited/discussed as suggested: DOI: 10.1371/journal.pone.0294769 DOI: 10.3389/fphar.2024.1369659 DOI: 10.1002/cbdv.202402548 please see the second last paragraph of the section Introduction ( page no 2) of the revised version of the manuscript. 4. Motivation for using multiview data and graph learning is justified but would benefit from a brief example of why single-view models may miss relevant interactions. Answer: As suggested by the reviewer we have now added an explicit example highlighting that single-view models, for instance relying solely on sequence similarity, may overlook biologically relevant interactions identified through functional or network-based similarities. Please see the second paragraph of section Introduction page no 3 line no. 65-77. 5. The integration of STRING, BioGRID, and sequence-based interactions is appreciated. However, the criteria for interaction confidence thresholding are not clearly explained. Were low-confidence interactions filtered? The authors can potentially benefir from https://doi.org/10.1021/acsomega.3c07866; https://doi.org/10.3389/fphar.2025.1509263 ; https://doi.org/10.3389/fphar.2025.1509263 ; https://doi.org/10.1007/s10989-020-10076-w Answer: Thank you for this point. We now clarify in the Methods section that only experimentally validated, high-confidence interactions were included from both CCSB and HPRD interactome resources, and low-confidence interactions (e.g., based on single sources or insufficient evidence) were filtered out during preprocessing. Interactions with ambiguous or duplicate entries were also removed to maximize data reliability. Additional references as suggested ([DOI: 10.1021/acsomega.3c07866], [DOI: 10.3389/fphar.2025.1509263], [DOI: 10.1007/s10989-020-10076-w]) have been considered and cited. Please see the last paragraph of the subsection ‘The Human Protein Interactome’ of the section ‘Method’ in the revised version of the manuscript (page no. 11-12, line no: 423-431). 6. The process for fusing multiview networks via Wasserstein distance is conceptually interesting, but the explanation is too terse. Please provide a schematic or algorithmic pseudocode to support reproducibility. Answer: We thank the reviewer for this suggestion. The schematic (Figure 1) workflow provides a detailed illustration of the entire workflow, including how the Wasserstein distance is employed to fuse the multiview network data. This figure visually demonstrates the integration of features from sequence similarity, gene ontology, and PPI networks using optimal transport. In addition, we have now added an explicit algorithmic pseudocode (see "Algorithm 1: Interaction prediction between SARS-CoV-2 and Human protein," in the section ‘Materials and Methods’ , page no. 14 of the revised version), which presents a clear and reproducible stepwise procedure. 7. The model is claimed to use GCN and Wasserstein integration, but details on the number of layers, training epochs, loss function, and hyperparameter tuning are missing. Answer: Detailed hyperparameters are now included: We used a three-layer GCN for each network, as empirically determined, trained for 50 epochs with the Adam optimizer (learning rate 0.001, dropout rate 0.1, activation: ReLU). Hyperparameter choices were validated by performance on validation/test splits (8:1:1), and additional ablation experiments (varying layers) are now discussed in the ‘Materials and Methods’ section. Please see the subsection ‘Extracting node features using GCN‘ page no. 13, line no 455-474 of the revised version of the manuscript. 8. It is unclear how negative samples were generated or balanced. Were negative protein pairs randomly sampled or drawn from known non-interacting pairs? Answer: Negative samples were generated by random sampling of protein pairs not present in the verified interaction set, ensuring they did not overlap with known positive interactions (see subsection “Extracting node features using GCN”, page no. 13, line no 462-474 of the revised version of the manuscript). This random negative sampling aligns with common practices in PPI prediction. 9. Figure 1 shows model architecture but lacks annotations and descriptions. It is hard to interpret the role of each module in the workflow. Answer: We have revised Figure-1 and its legend to include detailed annotations for each module, explicitly indicating the roles of GCN, multi-view integration, clustering, and interpretation modules, facilitating workflow comprehension. Please see figure-1 and legend in the revised version of the manuscript. 10. There is no table presenting the model’s quantitative performance (e.g., accuracy, F1-score, AUC) on benchmark datasets. This weakens the credibility of performance claims. Answer: Table 1 now summarizes the model’s quantitative performance on all three network types, including ROC-AUC and AP scores for validation and test sets. Please see the text in the subsection ‘ Comparative Evaluation with Baseline Methods’ in page no. 6, line no 184-198 of the revised version of the manuscript. 11. Add comparative results against baseline methods (e.g., logistic regression, DeepWalk, GCN without multiview fusion). Answer: We acknowledge the reviewer’s valuable suggestion. However, our proposed framework uniquely integrates embeddings from multiple views (PPI, GO, sequence similarity) through Wasserstein-distance-based fusion, rather than merely relying on single-view embeddings. To meaningfully highlight our model’s superiority, we have now added comparative analyses against the following baseline scenarios: Single-view GCNs individually applied on each network (PPI, GO, Sequence similarity). Simple fusion strategies (concatenation and averaging) of single-view GCN embeddings. Standard embedding techniques such as DeepWalk followed by a simple concatenation fusion method. Results demonstrate that our Wasserstein fusion approach significantly improves performance (see table 1). These additional analyses further validate the robustness and effectiveness of the proposed multi-view integration strategy. Please see the subsection ‘Comparative Evaluation with Baseline Methods’ in page no. 6, line no 184-198 of the revised version of the manuscript. 12. The identification of known host proteins (e.g., ACE2, TMPRSS2) adds confidence, but the analysis is anecdotal. Perform GO enrichment or pathway analysis on the predicted top 100 proteins to support biological relevance. Answer: We agree with the reviewers and have now performed GO enrichment analysis for the top 100 predicted proteins, highlighting significant pathways (e.g., viral entry, immune response). Results are provided in Supplementary Table-2 and briefly summarized in the main text. Please see the last paragraph of the subsection ‘Biological relevance and therapeutic potential of predicted host factor’ page no. 9 line no. 319-327 and supplementary table-2 of the revised version of the manuscript. 13. Drug repurposing suggestions (e.g., dexamethasone, baricitinib) are mentioned, but how were these inferred from the predicted PPIs? The method of linking predictions to drugs should be more systematically described. Answer: A systematic workflow was used: predicted host proteins were mapped to drug–protein associations using the CMAP Drug Repurposing Hub ([30]), identifying drugs with existing links to these proteins. The method is now described in detail in the second last paragraph of the subsection’ Predicted host factors promotes Host Directed Therapy (HDT) option against SARS-CoV-2’. Please see page no. 10, line no. 344-368 of the revised version of the manuscript. 14. The conclusion is too generic. Reflect more critically on the limitations, such as the dependence on existing PPI databases and lack of experimental validation. Answer: The Conclusion now explicitly discusses current limitations, including:dependence on available PPI databases,absence of direct experimental validation for new predictions, possible incompleteness of host interactome coverage. We stress that these limitations motivate future validation and model expansion. See the section 'Discussions’ of the updated version of the manuscript, page no. 18, line no. 652-668. 15. Future directions could include validation using CRISPR screens or cross-species PPI transfer learning. Answer: We appreciate this suggestion and agree that our predictions could be substantially strengthened by experimental validation, particularly using CRISPR-based genetic screens to confirm host factor essentiality. Additionally, cross-species PPI transfer learning represents a promising strategy to generalize our findings and discover conserved interaction patterns across related viral-host systems. We have now explicitly included these as important future research directions in the section 'Discussions’ of the updated version of the manuscript, page no. 18, line no. 662-671. 16. The manuscript needs a thorough language edit for grammar and clarity. Examples: Page 5: “This approach is very efficiently to...” → “This approach is efficient for...” Page 8: “The ROC curve indicate the...” → “The ROC curve indicates. Answer: The manuscript has undergone comprehensive language editing for grammar, clarity, and style. Examples such as “This approach is very efficiently to...” and “The ROC curve indicate the...” have been corrected as suggested. Answer to comments of Reviewer 2 Comments: 1. Clustering for proteins can be overlapping in nature in biological data, but the authors use the hierarchical clustering which detects disjoint clusters and has a high time computational cost and also the method is old and could be error prone. – Proper justification require. Answer: Thank you for highlighting this point. While hierarchical clustering indeed produces disjoint clusters and can be computationally intensive, we chose it primarily for its interpretability and wide adoption in bioinformatics, particularly for protein interaction data where distinct biological modules or functional groups are frequently analyzed. To ensure robustness, we employed Ward's linkage to minimize intra-cluster variance, facilitating the meaningful biological interpretation of identified protein clusters. Moreover, hierarchical clustering allowed us to clearly visualize and interpret the results through dendrograms and cluster-specific heatmaps, which are beneficial for biological insights. Nevertheless, we acknowledge that overlapping clustering methods, such as fuzzy clustering or community detection algorithms (e.g., Louvain, Infomap), could potentially provide more nuanced biological insights due to the inherently overlapping nature of protein functional associations. Future analyses could explore these advanced clustering methods for additional validation and biological insight. We have now mentioned this in the subsection ‘Hierarchical clustering of the Wasserstein distance matrix’ page no line no. 240-251 of the revised version of the manuscript. 2. Algorithm also has high overall time complexity so it would be slow for large datasets. –Need more statement about it. Answer: We agree and have clarified in the Methods (in subsection time complexity analysis) that while sequence similarity computation and hierarchical clustering contribute to a high theoretical time complexity (O(n²m² + n³)), our study’s dataset size allowed practical execution. For much larger datasets, steps such as pairwise similarity calculation and clustering could be optimized using parallelization or approximate algorithms. We now explicitly discuss this scalability limitation and possible future solutions in the manuscript’s Discussion section. Please see the section ‘Discussions’ in the revised version of the manuscript, page no. 18, line no. 652-660. 3. What is significance of you own algorithm in relation to other already available similar algorithms? A comparative analysis could be better for the paper and could enhance the novelty of the work. Answer: Thank you for this important point. The novelty of our approach lies in the integration of three complementary biological views (PPI, sequence similarity, GO similarity) via graph convolutional networks, and in the application of Wasserstein (optimal transport) distance for protein similarity, which collectively outperform single-view or conventional network methods. We have now included a comparative performance analysis against baseline algorithms—logistic regression, DeepWalk, and single-view GCN—in the Results section (see subsection ‘Comparative Evaluation with Baseline Methods’ table-1, page no.06 line no. 184-198). Our multi-view GCN-Wasserstein approach achieved higher ROC-AUC and average precision scores, demonstrating its superiority in both prediction accuracy and biological interpretability. 4. These proteins already serve the purposes of host-directed therapy (HDT) options for other viruses, thereby also being potential candidates for building HDT strategy against SARS-CoV2. --- “thereby also being potential candidates for building HDT”—More explanation require. Answer: We have expanded the explanation. Host-directed therapy (HDT) aims to inhibit host proteins that viruses exploit for infection and replication. Many of our predicted host factors are already validated as essential in the life cycle of multiple viruses. Therefore, these proteins are not only relevant to SARS-CoV-2 but are also proven targets in existing HDT strategies for other viral infections. By targeting such proteins, repurposed drugs have the potential for broad-spectrum antiviral effects and reduced risk of resistance compared to virus-targeted therapies. We have now included an analysis to further evaluate the biological relevance of our predicted proteins. We have performed Gene Ontology (GO) enrichment analysis on the top 100 predicted human proteins (host factor) it is observed that significantly enriched terms include processes such as viral entry into host cell (GO:0046718, $p = 3.1 \times 10^{-6}$), regulation o Attachments Attachment Submitted filename: Reviewers Comments_plosone (1).pdf https://doi.org/10.1371/journal.pone.0332794.r002
4 Sep 2025 Decision Letter - Chandrabose Selvaraj, Editor A Graph Neural Network-Based Approach for Predicting SARS-CoV-2–Human Protein Interactions from Multiview Data PONE-D-25-21081R1 Dear Dr. Ray, We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements. Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication. An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact billing support. If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org. Kind regards, Chandrabose Selvaraj, Ph.D. Academic Editor PLOS ONE Additional Editor Comments (optional): Reviewer #1: Reviewer #2: Reviewers' comments: Reviewer's Responses to Questions Comments to the Author 1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation. Reviewer #1: All comments have been addressed Reviewer #2: All comments have been addressed ******** 2. Is the manuscript technically sound, and do the data support the conclusions? The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented. Reviewer #1: Yes Reviewer #2: Yes ****** 3. Has the statistical analysis been performed appropriately and rigorously? Reviewer #1: Yes Reviewer #2: (No Response) ****** 4. Have the authors made all data underlying the findings in their manuscript fully available? The PLOS Data policy requires authors to make all data 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 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—e.g. participant privacy or use of data from a third party—those must be specified. Reviewer #1: (No Response) Reviewer #2: Yes ****** 5. Is the manuscript presented in an intelligible fashion and written in standard English? PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here. Reviewer #1: (No Response) Reviewer #2: Yes ****** 6. Review Comments to the Author Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters) Reviewer #1: The authors have addressed my comments. The manuscript is accepted for publication. I have no further comments. Reviewer #2: (No Response) ****** 7. 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 ******** https://doi.org/10.1371/journal.pone.0332794.r003
Formally Accepted
Acceptance Letter - Chandrabose Selvaraj, Editor PONE-D-25-21081R1 PLOS ONE Dear Dr. Ray, I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now being handed over to our production team. At this stage, our production department will prepare your paper for publication. This includes ensuring the following: * All references, tables, and figures are properly cited * All relevant supporting information is included in the manuscript submission, * There are no issues that prevent the paper from being properly typeset You will receive further instructions from the production team, including instructions on how to review your proof when it is ready. Please keep in mind that we are working through a large volume of accepted articles, so please give us a few days to review your paper and let you know the next and final steps. Lastly, if your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org. You will receive an invoice from PLOS for your publication fee after your manuscript has reached the completed accept phase. If you receive an email requesting payment before acceptance or for any other service, this may be a phishing scheme. Learn how to identify phishing emails and protect your accounts at https://explore.plos.org/phishing. If we can help with anything else, please email us at customercare@plos.org. Thank you for submitting your work to PLOS ONE and supporting open access. Kind regards, PLOS ONE Editorial Office Staff on behalf of Dr. Chandrabose Selvaraj Academic Editor PLOS ONE https://doi.org/10.1371/journal.pone.0332794.r004

Open letter on the publication of peer review reports

PLOS recognizes the benefits of transparency in the peer review process. Therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. Reviewers remain anonymous, unless they choose to reveal their names.

We encourage other journals to join us in this initiative. We hope that our action inspires the community, including researchers, research funders, and research institutions, to recognize the benefits of published peer review reports for all parts of the research system.

Learn more at ASAPbio .