Dear Dr Ribeiro,

Thank you very much for submitting your manuscript "Exploring the Potential of Structure-Based Deep Learning Approaches for T cell Receptor Design" 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,

Dina Schneidman

Academic Editor

PLOS Computational Biology

Nir Ben-Tal

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: In this work, the authors evaluate different inverse folding algorithms for TCR-pMHC binding design.

This is great and thorough work. i have a few comments.

- you use sequence recovery as a main metric. Given that one wants diversity in the designed sequences, why is sequence recovery, which measures the conservation of native positions, a good metric? can you explain this better in the manuscript?

- can you also use immunebuilder (in additon to tcrmodeler2) to investigate the designs structurally?

- can you also run your designed TCRs through OLGA and test if they are actually actually native-like TCRs? If the sequence pgen is very low, then the designs are likely not very natural and thus not useful for therapeutics application (immunogenic).

Reviewer #2: In this manuscript, Ribeiro-Filho et al. explore the application of inverse folding models (ProteinMPNN and ESM-IF1) to the fixed-backbone CDR sequence design of TCRs. They evaluate the models’ performance via sequence recovery, refolding, and energetic scoring (using Rosetta and MM/PBSA calculations). The authors demonstrate that the tested inverse folding methods can recapitulate a substantial portion of native sequences and produce designs with equal, and in some cases greater, predicted binding affinity to the target peptide-MHC complex than the starting TCR sequence.

This work provides value to the field by examining the performance of inverse folding models on TCRs, which, as the authors point out, are underexplored in the protein design space. However, the manuscript is currently somewhat limited in novelty and scope. The authors use existing, well-established inverse folding models and evaluation methods. Additionally, the sequence design analysis is primarily limited to CDR3 loops. The paper would be strengthened by addressing the following points.

Major comments

• The manuscript should make a stronger case for the intended application(s) and the applicability (or lack thereof) of the inverse folding models. As it stands, only the re-design of CDR (primarily CDR3) loops for structurally-resolved TCR-pMHC complexes is investigated. This may only cover a fraction of real-world use cases for TCR inverse folding. Extending the analyses to some of the following suggestions would provide a more thorough understanding of the models’ applications and limitations:

– What are the inverse folding models stronger or weaker at, respectively and/or in general? A more detailed investigation of failure modes would be helpful.

– Expand sequence design beyond the CDR3. Given that the randomly designed CDR3 sequences had similar RMSDs to the designed sequences and TCRs with the same CDR3 sequence can bind different pMHC complexes (as mentioned in section 4.1), an expanded analysis of sequence design for other TCR regions could provide more insights into model performance. Additionally, this would open the door to further important evaluations, such as whether generated sequences map back to known V/J genes (and, if so, whether these genes are associated with the native loop canonical forms).

– Apply the inverse folding approaches to predicted structure inputs. In real design use cases, one may not always have a solved structure of the starting complex. Are the models still able to achieve strong performance without an experimentally solved structure?

– Polyreactivity: although the authors state this is beyond the scope of their manuscript, it would be valuable to understand, for example, if the inverse folding model suggestions increase affinity for the MHC rather than the peptide and thereby increase the risk of polyreactivity.

• The MM-PBSA affinity calculation benchmark gives a helpful understanding of the potential accuracy of this approach. However, this accuracy is not directly applicable to the use case in this manuscript: for the benchmark, only cases with a solved structure of the wild-type and mutant were included; for the assessment of inverse folding designs, the mutant structures were modeled. The MM-PBSA benchmark should be updated/extended to include cases with modeled mutant structures to provide a relevant quantification of this protocol’s accuracy.

• It would be useful to provide an understanding of natural sequence variability in TCRs across different portions of the structure (down to single-position resolution for CDR3 at least). This will give more insights into the relative challenges of sequence recovery tasks.

Minor comments

• It would be helpful if the authors more clearly stated/explained the following:

– In section 2.1, it should be more clearly stated that the antigen is provided as context for the models. Additionally, the authors should explain whether the rest of the TCR sequence is given to the model as context and whether the CDR sequences were sampled autoregressively or in a single forward pass.

– In Figure 1B, a sequence recovery value of exactly 50% appears to be overrepresented. Is there an explanation for this?

– What percent of model designs are redundant?

– In the Figure 4 legend, the buried AAs are defined as a subset of the hotspot AAs, but this does not seem to match section 4.5.

– In section 2.5, the TCRModel2 confidence metric should be explained, as this is a major underpinning of the analyses presented in this section. Additionally, it should be clearly specified which confidence metric is used. (I presume the linear combination of pTM and ipTM, as this is defined as ‘model confidence’ in the TCRModel2 publication, but the paper also states that pLDDT, pTM, and ipTM confidence metrics are provided).

– Also in section 2.5, the authors should state and comment on the results in Figure S10, that random CDR3 sequences achieve similar RMSDs to designed sequences.

– Could the authors elaborate on how the same sequence can achieve different ProteinMPNN scores, as stated in the Figure S5 caption? Would the log-probability outputs not be fixed for a given structure input (and used to sample/score multiple sequence outputs)?

• The following text could be updated:

– In the abstract, I would suggest that immunotherapeutic applications of design approaches are "underexplored" rather than "unexplored".

– In the introduction, I believe that it is not fully accurate to say that ProteinMPNN designs “entirely novel proteins”, as the backbone is not novel (fixed). Perhaps this could be amended to “novel sequences”.

– In section 2.1, a more likely explanation for the poorer performance on MHC-II- than MHC-I-binding TCRs is the lower number of training structures, not test structures. There does not seem to be any evidence to support that the test cases included here would be more challenging than other potential MHC-II test cases.

– Towards the end of section 2.1, “interestingly” should be removed. As the authors point out, sequence recovery is expected to be higher across all rather than just interface CDR3 positions due to greater sequence conservation.

– The last sentence of section 2.4 would require more supporting evidence. The data could alternatively be interpreted as the sequence recovery remaining unchanged due to the relatively low/moderate recovery success; the models could, for example, only be recovering non-interface or non-essential binding positions.

• The ESM-IF1 inverse folding model should be referred to as “ESM-IF1” rather than “ESM-IF”.

• The figure colors should be kept consistent throughout (e.g., always green for native).

• In Figure 2B, one CDR3b appears to be in a very different place than the other CDRs. Could the authors comment on this?

• The amino acid groupings in section 4.4 are missing P and T. Additionally, it would be helpful to keep the groupings consistent with Figure 3.

• Figure S12B should be square (equal x- and y-axes) to show that, although the correlation is strong, the calculated and experimental dG results are not on exactly the same scale.

• The caption of Figure S4 only explains A-B, but the figure includes panels A-D.

**********

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

**********

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: Yes: Alissa M Hummer

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 Ribeiro,

Thank you very much for submitting your manuscript "Exploring the Potential of Structure-Based Deep Learning Approaches for T cell Receptor Design" 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,

Dina Schneidman

Academic Editor

PLOS Computational Biology

Nir Ben-Tal

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: The authors have addressed all of my comments.

Reviewer #2: The authors have made significant revisions to the manuscript, addressing the feedback from the reviewer comments and strengthening the paper. The added analyses provide a clearer understanding of the strengths, limitations and functionalities of inverse folding models for TCR sequence design. Additionally, the updates to the Discussion section highlight interesting points and areas for future exploration.

I have a few minor comments arising from the changes. The line numbers are indicated for the version with tracked changes.

Minor analysis

• Line 171, Fig S7: Are there any unifying features for the cases which ESM-IF1 or ProteinMPNN, respectively, performed better on? (i.e. those which are above vs. below y=x in Fig S7)

• Fig 5: Could you include a figure and statistical analysis similar to Fig 5B for the information portrayed in Fig 5A? (i.e. Fig 5B but with 'maximum sequence recovery' rather than 'maximum sequence recovery difference')

• Line 512, Fig S27: Can you identify any explanations (e.g., specific sequence changes) for why designing all CDRs substantially improved binding affinity as compared to designing only CDR3 for 7pbe?

Text – clarification

• Line 352: Could this new text be rephrased for clarity? The statements in lines 349 and 352 seem contradictory at a first glance (the differences being analyzed in these sentences – with vs. without the pMHC; ProteinMPNN vs. ESM-IF1 – could be explained more clearly). The sentences in lines 354-357 also appear contradictory; these concepts are more clearly explained in the response to reviewers comment.

Text – minor changes

• Line 24: The phrase "trained on" in "this study explores whether computational methods, trained on deep learning architectures such as ProteinMPNN and ESM-IF1" is misleading and should be amended to clarify that ProteinMPNN and ESM-IF1 are directly used (as opposed to new methods trained/based on these architectures).

• Line 336: It would be helpful if more information was directly provided in the Results section about the method for identifying hotspots (using Rosetta, 0.5 kcal/mol cutoff) and the relevant methods section (4.7) was referenced.

• A few typos are included and should be fixed, e.g., "residues" to "resides" in line 39 and "10-10" to "10^-10" in line 296.

Figures – minor changes

• Figure S3C – The sequence position numbers are absent from panel C (included in A, B and D).

**********

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: No: 

Reviewer #2: None

**********

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: Yes: Alissa M. Hummer

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 Ribeiro,

We are pleased to inform you that your manuscript 'Exploring the Potential of Structure-Based Deep Learning Approaches for T cell Receptor Design' 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,

Dina Schneidman

Academic Editor

PLOS Computational Biology

Nir Ben-Tal

Section Editor

PLOS Computational Biology

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