Dear Dr. Guo,

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 Apr 24 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.

• 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,

Tao Huang

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. Thank you for stating the following financial disclosure:

       1. The National Key Research and Development Program of China (2022YFC2603305)

       2. Key Research and Development Program of Yunnan (202103AQ100002)

       3. Central Funds Guiding the Local Science and Technology Development (202207AB110017)

       4. Yunnan Academician and Expert Workstation (202205 AF150023)

       5. Yunnan Fundamental Research Projects (202201AY070001-192)

       6. The Scientific and Technological Innovation Team in Kunming Medical University (CXTD202215)

Please state what role the funders took in the study. If the funders had no role, please state: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

If this statement is not correct you must amend it as needed.

Please include this amended Role of Funder statement in your cover letter; we will change the online submission form on your behalf.

3. Please include a separate caption for each figure in your manuscript.

4. 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.

[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?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #5: No

Reviewer #6: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously? -->?>

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #5: No

Reviewer #6: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available??>

The PLOS Data policy

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #5: Yes

Reviewer #6: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English??>

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #5: Yes

Reviewer #6: Yes

**********

Reviewer #1: The english text sounds well in all paper sections, the topic and idea of the study original and absolutely current. Methodology applied is correct as well as statistic analysis.

I only suggest to better summarize and focus the introduction around the study question.

Reviewer #2: AD-GCN: a novel graph convolutional network integrating multi-omics data for enhanced Alzheimer's disease diagnosis, the manuscript is interesting, it is generally well written, with some grammatical errors. The manuscript is generally well-structured, with each section logically progressing from one topic to the next

Abstract

1. Line 37-38: multi-omics risk factors, such as age, sex, education level, and genomic and epigenomic influences. This should be changed, as when I think of multi-omics I think of genomics, transcriptomics etc, and not of age, sex, etc.

Introduction

1. The authors should include a small introduction about what AD is. An epidemiology of AD in China if this information is possible.

2.Line 96: Contributions should be changed to aims.

Methods

1. Line 108. Remove from the ADNI.

2. What are the 12 cognitive function scores? If possible and this information is available the tests used should be added.

3. The ROSMAP dataset was also obtained from ADNI or from somewhere else? as it is not clear too me.

4. Line 126-127: It would be better to represent the following as a table.

5. What tool was used for statistical analysis? R, excel, SPSS?

6. Line 177: What are the tree ML integration strategies that were implemented?

7. Was previously work done in order to decide the five ML methods that would be applied?

Results

1. Add full names of the cognitive tests used under the table.

2. 1KGP cohorts, why was this cohort not mentioned in the methods? unless I have missed it somewhere.

3. A justification for the threshold values used in the QC process (why specific minor allele frequency or Hardy-Weinberg deviations thresholds were selected) would help readers understand the rationale for these decisions.

4. The comparison between the four PRS models (P+T, LDpred, PRS-CS, and lassosum) is clear. However, a bit more detail on the algorithmic differences of each model would benefit readers who may not be familiar with these methods. For instance, what makes LDpred particularly suitable for this kind of analysis compared to PRS-CS or P+T?

5. AUC analysis is very informative, but it would be helpful to state if improvements are statistically significant (or not), and whether the differences between models (e.g., LDpred vs. lassosum) are meaningful in a clinical context. Adding some statistical tests (e.g., pairwise comparison of AUCs) would provide stronger evidence.

6. The results suggest that PRS models are particularly useful in distinguishing CN from AD, but the manuscript could benefit from a deeper exploration of why this is the case.

7. The authors report KS values and p-values for the linear fits based on LDpred PRS, but a more thorough interpretation of the statistical significance in the context of AD stages would be helpful.

8. The description of methylation data processing using ChAMP and feature selection via Fisher Score and entropy filtering is rigorous, but I would encourage the authors to provide more detail on the thresholds used for quartile-based feature selection. How were these thresholds determined, and do the results vary with different thresholds? This will help readers assess the robustness of your feature selection approach.

9. The AD-GCN as a novel approach for multi-omics integration is particularly interesting. However, the technical description of the graph convolutional network (GCN) could benefit from additional explanation. Specifically, how are cognitive scores incorporated into the graph model? Are these scores treated as node features or edge weights? Further clarification on how the cognitive scores are integrated into the graph structure would help readers better understand this innovative aspect of the model.

Discussion

1. The authors should mention the benefits of this approach to AD and how it may be applied to other NDs as well, as this approach is novel and holds great research potential .

2. The authors should mention the limitations of their work.

General comments.

1. Add full name followed by abbreviation. e.g. line 99.

2. Add full name and links of databases. Also add references to the the database if there is.

3. Sample Homogeneity: The use of only Caucasian participants is understandable, but there should be a more thorough discussion of potential limitations this introduces, particularly in terms of generalizability to other ethnic groups. Given that AD and related diseases have genetic heterogeneity across populations.

4. The figures should be made into a higher resolution.

Reviewer #3: This study presents AD-GCN, a novel graph convolutional network integrating multi-omics data (clinical, genomic, and methylation) to enhance Alzheimer’s disease (AD) diagnosis. The authors demonstrate that multi-omics integration outperforms single-omics approaches and that AD-GCN surpasses traditional machine learning (ML) methods. While the work is innovative and addresses a critical need in AD research, several methodological and presentation issues require clarification and improvement.

Strengths:

�The idea of using graph convolutional networks to integrate multi-omics data for AD diagnosis is novel and has the potential to capture complex interactions among different data types.

�The use of polygenic risk scores and feature selection methods is well justified and shows promising results in improving classification performance.

�The experiments are well designed, with comparisons between single-omics and multi-omics models, as well as ablation studies to evaluate the contributions of different components in the AD-GCN model.

�The results demonstrate that the AD-GCN model outperforms traditional machine learning ensemble methods, highlighting its potential clinical utility.

Weaknesses:

�The study is limited to Caucasian individuals, which may limit the generalizability of the findings to other racial groups. Broader inclusion of diverse populations would strengthen the study.

�The availability of public datasets is acknowledged as a limitation. The ADNI cohort includes only 35 AD cases, which raises concerns about statistical power and generalizability. Expanding the number of accessible datasets could further validate and improve the model.

�The biological interpretation of the selected methylation markers could be strengthened with additional functional validation studies.

Specific Comments:

1、Abbreviations (e.g., BDI, MCC) should be defined at first mention.

2、The Introduction section provides a comprehensive background on AD and multi-omics data integration. However, more details on existing graph convolutional network approaches for similar applications could be included to position the current work more clearly within the field.

3、LDpred outperforms other PRS methods, but the manuscript lacks a discussion of why LDpred is better suited for AD risk modeling compared to alternatives like PRS-CS.

4、Validate Generalizability: Test AD-GCN on ROSMAP or another independent cohort.

5、The Discussion overemphasizes technical contributions; a deeper exploration of biological insights and future research directions could be further expanded. .

Recommendation:

Overall, this is an interesting and potentially impactful study. With some minor revisions to address the limitations and clarify certain aspects of the methods, I recommend this manuscript for publication in PLOS ONE after minor revisions. The novel approach and promising results indicate the potential for significant advances in AD diagnosis and personalized medicine.

Reviewer #4: The study presents a novel graph convolutional network (GCN) approach for integrating clinical, genetic, and methylation data, offering a promising method for AD diagnosis. However, a few refinements could further strengthen the manuscript:

1. Clarity in Structure – The Results and Methods sections are interwoven, making it challenging to follow the study’s flow. A clearer separation of these sections would improve readability and comprehension.

2. GWAS and PRS Methodology – The description of GWAS and polygenic risk score (PRS) analysis needs more clarity. Instead of performing GWAS on a relatively small dataset, the authors could consider leveraging publicly available GWAS summary statistics to enhance the robustness and interpretability of findings. Similarly, more details on the thresholds and criteria used in PRS calculations and random forest feature selection would improve reproducibility.

3. Clinical Applicability – The Discussion section could be expanded to address how AD-GCN could be integrated into clinical practice and what potential limitations or challenges might arise in real-world applications.

4. Some sentences are not very fluent and the figures are not very clear.

These refinements would help better highlight the study’s contributions and improve its overall impact.

Reviewer #5: The authors developed a machine learning tool named AD-GCN for Alzheimer's disease diagnosis using clinical data, genomic variants, and DNA methylation data. The biggest issue is there are only 35 AD cases that have all three types of omics data, such a small sample size is not enough for the model to extract features. In addition, CN cases were 6 times more than AD cases, and MCI cases were 10 times more than AD cases in the ADNI dataset. The sample size imbalance in these groups will cause super-control bias in the model prediction. The average AUC of all 5 methods in MCI vs. AD and CN vs. AD are lower than 0.75, the AUC of two methods in CN vs. AD even fluctuates around 0.5 (Figure 5). This means that the above models are useless for AD diagnosis. Although the author did many analyses and evaluations of these models, the insufficient sample size is the fatal flaw of the manuscript.

Reviewer #6: The manuscript presents a well-executed study on integrating multi-omics data for Alzheimer’s disease classification using a novel graph convolutional network (AD-GCN). The work is methodologically strong, and the results are promising. The comparisons with traditional machine learning models highlight the advantages of AD-GCN in handling complex multi-omics relationships.

Strengths:

• The use of a graph convolutional network (GCN) to integrate omics data is an interesting and innovative approach.

• The comparison between single-omics and multi-omics classification methods is well executed and demonstrates the benefits of data integration.

• The inclusion of polygenic risk scores and methylation features strengthens the study’s ability to capture genetic and epigenetic risk factors for AD.

Areas for Improvement:

• Biological Interpretation: The study identifies key genetic (APOE, CR1, TOMM40) and epigenetic (AACS, VTRNA2-1, CYP26C1) markers, but their roles in AD pathogenesis are not discussed in depth. A more detailed interpretation would be valuable.

• Justification for Adjacency Matrix Construction: The authors use sample similarity and cognitive scores to build the adjacency matrix in AD-GCN. Could they clarify why this approach was chosen over other possible graph structures (e.g., biological pathway-driven networks)?

• Computational Efficiency: AD-GCN is shown to outperform traditional ML models in classification accuracy, but the manuscript does not discuss its computational requirements. Providing details on runtime, memory usage, and scalability would be helpful for researchers looking to apply this method.

• Generalizability and Population Diversity: The study is based on ADNI and ROSMAP, which primarily include Caucasian participants. Have the authors considered validating the model on more diverse datasets?

• Reproducibility: The authors mention that their code is available on GitHub. Could they confirm whether preprocessing scripts and detailed environment configurations are included to allow full replication of results?

Final Recommendation:

This manuscript presents a valuable contribution to computational approaches for AD diagnosis and is well suited for publication after minor revisions. Addressing the interpretability of genetic findings, discussing computational feasibility, and clarifying the adjacency matrix construction would further strengthen the study.

**********

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

Reviewer #4: No

Reviewer #5: No

Reviewer #6: Yes:  Ece Eldem

**********

[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.

AD-GCN: A novel graph convolutional network integrating multi-omics data for enhanced Alzheimer's disease diagnosis

PONE-D-25-05441R1

Dear Dr. Guo,

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. If you have any questions relating to publication charges, please contact our Author Billing department directly at authorbilling@plos.org.

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,

Tao Huang

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #2: (No Response)

Reviewer #3: All comments have been addressed

Reviewer #6: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions??>

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #6: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously? -->?>

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #6: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available??>

The PLOS Data policy

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #6: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English??>

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #6: Yes

**********

Reviewer #2: I thank the authors for the changes made to the manuscript. I have no further comments or concerns to add and I am happy with the revisions made.

Reviewer #3: (No Response)

Reviewer #6: I appreciate the authors’ thorough and thoughtful revisions. The updated manuscript has addressed the concerns raised in the previous round. The manuscript now reads more clearly, and the methodological explanations are better justified. I find the revised version technically sound and suitable for publication.

**********

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 #2: No

Reviewer #3: No

Reviewer #6: No

**********