Application of Machine Learning Algorithms for Prediction of Tumor T-Cell Immunogens
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe manuscript presented here describes a review of major approaches to identify immunogenic tumor antigens across their datasets and post validation, they further incorporated the best performing methods on their server.
Since the paper is more of a review of the available prediction techniques, it would be nice if the authors add a few words on the basic principle behind each of the available models.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThis is an awesome manuscript introducing ML application for the prediction of tumor antigens. The introduction, method, and results are well-written, and are explained very clearly for the readers to understand, even for someone who is not familiar with ML. The identified important features are very interesting, and I would hope to see more interpretation of these features for the audience to get some mechanistic understanding on why these features might be more important than others. Any discussions or citations will be really interesting and helpful.
Comments:
- Line 187-188: For equation (1) and (2). Is it possible to use a letter instead of “lag” to represent lag-value? Just for aesthetic reason.
- Line 190 - Is there a typo? j, k are indices not the E-descriptor? Also in later part of the manuscript terminology like ACC145 is used. It would be great to make these consistent for reader’s understanding.
- I might miss this in the manuscript, but have the authors tried feature selection or only use the features identified as important as importance ranking suggest? It would be great to know if any model with fewer-than 175 features can achieve comparable performance.
English is good and I have no problem understanding it. I will leave the editor for any minor language improvement.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsManuscript ID: applsci-2904100
Title: Application of Machine Learning Algorithms for Prediction of 2 Tumor Antigens
The manuscript described identification of tumor antigens by different tools of machine learning methods.
Reviewer suggested comments mentioned below-
Manuscript has been written very poorly and fills with lot of grammatical mistakes. Scientific soundness of sentences is very low. Authors need to revise each section of the manuscript thoroughly; in many places sentences are not clear and grammatically incorrect. The written manuscript has many places formatting issues. Check it and revise manuscript.
Author’s affiliation should be recheck and provide complete information?
Abstract must be improvised and written grammatically correct. Sentences should be written short and self-sufficient to understand.
In Line 8-9: Sentence must be recheck and correct it (Cancer Vaccines is not appropriately written in the sentence).
In Line 13: Candidate vaccines should be vaccine candidates. Recheck and correct it.
References must be rechecked in whole manuscript, is that appropriate for the mentioned text?
Conclusion should be conclusive and in conclusion section there is no conclusive summary mentioned which can explain your study.
Line 34: In place of CAR-T therapy, it should be CAR therapy. Recheck and correct it
Line 37-38: “When this complex is recognized by T-lymphocytes, an immune response is initiated, rendering the antigen immunogenic.” What the means of written sentence? Check it and correct it
Title of Table 1 is correct? Recheck
In introduction, author explained the details of each Machine-learning tools. It must be revised and make separate title for better clarification in a separate heading.
Comments on the Quality of English Language
The English language is not up mark for scientific publication.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
Comments and Suggestions for AuthorsI congratulate the authors on this work. One point they may wish to consider is to briefly discuss or cite other studies conducted on the same topic, or to otherwise briefly highlight the novel aspects this study brings to the existing literature.
- It is a comparative evaluation of six ML algorithms to identify and predict immunogenic tumor antigens. In silico prediction algorithms, specifically those employing machine learning (ML) approaches, play a pivotal role in the accurate identification and characterization of immunogenic tumor antigens, as it is impractical to isolate and evaluate each putative antigen individually. These algorithms effectively curtail the extent of experimental work required for the discovery of candidate vaccines. The authors applied six supervised ML methods developed for the purpose of immunogenicity prediction: k nearest neighbor (kNN), linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), support vector machine (SVM), random forest (RF), and extreme gradient boosting (XGBoost), to a dataset comprising 212 experimentally validated human tumor peptide antigens and an equal number of non-antigenic human peptides. The models underwent rigorous validation through internal cross-validation within 10 groups from the training set and were further assessed using an external test set. The kNN model demonstrated superior performance, recognizing 90% of known immunogens in the test set. The RF model excelled in the identification of non-immunogens, accurately classifying 93% of them in the test set. The three top-performing ML models according to multiple evaluation metrics (SVM, RF, and XGBoost) are to be subsequently integrated into the new version of the Vaxi-Jen server, facilitating tumor antigen prediction through a majority voting mechanism.
- To my knowledge, this comparative evaluation has not been performed before. The authors need to empahasise the novel elements of this study by adding a brief statement to that effect in the study.
- A statement or table delineating the advantages this comparative evaluation brought comnpared with previous similar comparisons, if any.
- The references are appropriate, however, the paper would benefit for highlighting the novel aspects of this study. I suggest tabulation of all relevant studies that have used reviewed this topic, if any, and addition to the manuscript (if not present already).
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsManuscript ID: applsci-2904100
Title: Application of Machine Learning Algorithms for Prediction of Tumor Antigens
Reviewer suggested comments mentioned below-
Authors need to revise each section of the manuscript thoroughly; in many places’ sentences are not clear and grammatically not sound correct. Check it and revise the manuscript. The mentioned all the references should be in same format?? Recheck and is it all references appropriately sited inside the manuscript??.
The written manuscript primarily focused on T cell mediated recognition of tumor antigens by Machine learning tools, but not other immune cells population covered. Authors should be recheck of manuscript title is appropriate according to focused study??
Line 28: The immune system?? Check it and revise
Line 33-34: In Cancer immunotherapy not only CAR-T immunotherapy applied but in addition CAR-NK and CAR-Macrophage immunotherapy now validating against tumors. Better to use CAR immunotherapy in place of CAT-T??
Line 87-89: The written sentence is not clear and understandable. Check it and revise it. Better to make two sentences.
Line 115 & 131: In place of MHC-I and MHC Class I, should be written MHC Class-I and MHC Class-I respectively?? check it and revise.
Line 175: Word formatting needs to be checked. Similar? It should be “similar” in that place of sentence?? Check it and revise
Line 10, 483, 494, & 511: in silico & in vivo should be written in italics?? Check it and revise it
Figure legend 2: Describe more about the resulted figure in legend for better clarification??
Figure 1 & Figure 2: check formatting of title figures?? It should be in same format
Comments for author File: Comments.pdf
English language can be improvised
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
