Dear Dr. Meiler,

Thank you very much for submitting your manuscript "Predicting the Functional Impact of KCNQ1 Variants with Artificial Neural Networks" 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,

Joanna Slusky, Ph.D.

Associate Editor

PLOS Computational Biology

Arne Elofsson

Deputy 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: The manuscript by Phul et al describes the development of an ANN to predict functional consequences of variants in the KCNQ1 ion channel. It is well written and mostly easy to follow. The results are clearly described and I only have few reservations (see below). Most importantly, I would like to understand a bit better the authors’ negative dataset and how this influences the practical utility of the method.

Major

1.

The authors state in the abstract that the functional predictions “can assist physicians in taking appropriate treatment decision for patients with genetic heart rhythm disorder.” This may well be true, but it’s hard to assess given the information in the paper. Importantly (as the authors are aware), clinical utility requires more than accurate predictions, including that they are actionable. I believe KCNQ1 is, but that is not clear from the paper. In addition, the prediction methods would need to be included in assigning variant effects, and again it is unclear whether the methods are better than what’s out there. Thus, while I don’t doubt that the methods developed are both interesting and could become useful, I would ask that the authors provide a bit more detail how these methods can actually help in treatment or remove this from the abstract. Although only a wording issue, I have added this in “major comments” section as I think it is important.

2.

I am not familiar with the genetics of LQTS/LQT1, so it would be great if the authors could comment on whether all the disease variants are dominant, recessive, work via compound heterozygosity or what the genetics is. In particular, I would like them to provide evidence that training a model based on experimental data of homozygous cells is relevant for the disease (in particular since I assume KCNQ1 is a tetramer).

3.

It would be useful if the authors could explain better both the idea behind their “silent” variants and how they control for that they don’t lead to artificial improvements. I assume the problem is that there are not sufficiently many known benign variants (or variants that have “neutral” phenotypes according to the criteria in Table 1). I couldn’t easily find the numbers, but would be good to know exactly how many “real” “neutral/benign” variants they have. Also, given that many of the features would be zero (I think) for these silent variants, how do the authors know that the ANN isn’t just learning to recognize this? Does the method work as well to recognize real neutral variants as it does for the artificial ones? In this regard, I am curious about the statement:

“We observed that introduction of second hidden layer improved the performance of the models on silent mutations with no effect on the prediction accuracy for functional variants.” Ideally, the authors should provide some assessment of accuracy for actual benign variants even if the method is trained on the artificial ones also.

4.

If I understand correctly, in their predictions, the authors lump together loss of function and gain of function variants as one class. Is that correct? It would have been useful to know how many are in each class, and how the predictions work for them separately? I would e.g. have guess that the evolution-based methods work better for LOF than for GOF.

5.

The authors use a relatively simple PSSM-based method to estimate variant effects from the alignments. That’s of course completely valid, but as they are well aware there are plenty of more advanced methods out there that use more information in the alignments. I would recommend that the authors either use or benchmark with these methods or alternatively tone down the fact that the structure/biophysics-based methods are “better”. Perhaps just say that they are better than using a PSSM, but that more advanced sequence-based methods are known to provide better predictions than PSSMs (at least in many cases).

6.

The authors write “MCC values were computed by using Biochemical Library (BCL) wherein the threshold was adjusted individually for IKs, V1/2, τact and τdeact from 0 to 1 to achieve the best MCC for each phenotype.”. How precisely were these hyper-parameters chosen and did the authors do anything to avoid overfitting?

7.

One of the main points from the authors is that the predictions based on biophysical features can be used to gain mechanistic insight. It would this be useful and interesting with some examples of this. The input-sensitivity analysis provides a bit of this, but it’s not really very detailed. This is complicated further by the fact that many of these scores may be correlated with each other, and given that they use relatively many features could make it difficult to pinpoint specific effects. That said, I think it would be relevant if the authors could make this part a bit clearer and perhaps discuss some examples of variants that they think they can explain why cause LOF.

8.

I thank the authors for making a web-server available. That’s great. I would recommend that they also make their code available so that others in principle can look underneath the hood. Also, it would be good to store the multiple sequence alignment that is used to calculate the PSSM.

Minor:

p. 4, line 70: The authors write “membrane proteins” but refer only to data for KCNQ1 in LQT1. I suggest either making it clear that it’s only a single protein or provide references for other proteins whose destabilization and mistrafficking is involved in LQT1.

p. 4, line 81–82: The authors write “but these tools have limited applicability for KCNQ1”. Could they provide a citation or evidence for this?

p. 5, line 99: I’ll let it be completely up to the authors, but I would recommend deletion of “advanced and innovative”.

p. 13: The authors write “We found that three out of 14 features used for biophysical model failed to improve the predictions on the unseen dataset, therefore were excluded”, but I couldn’t see this analyses.

Reviewer #2: In the present manuscript, the authors used machine learning to develop an advanced method for functional prediction of KCNQ1 mutants. The functional classification categories of this method were then expanded by a scheme that was able to predict variant-specific changes in four electrophysiological KCNQ1 parameters (namely peak current density, voltage of half-maximal activation, and activation and deactivation time constants). The authors then evaluated the performance of artificial neural networks trained on evolutionary and biophysical features for KCNQ1 and observed that a combination of both features produced a model with good predictive accuracy.

This study is complementary to several previous studies in which artificial intelligence was used to predict protein structure and the properties of its mutants.

The concept behind this paper have merit. In fact, this concept is intriguing and has the potential to provide important new information for both basic (mode of action) and applied science (therapy).

Although this manuscript is original and novel, one major issue is that in this study, only a single type of protein is being tested as a case study which questions the generalization of the results. The authors should therefore test an additional type of protein to make the results and conclusions more universal.

In addition, the authors trained and also validated their models with 125 KCNQ1 variants that were functionally evaluated by others; there is thus a need for more functional data – for example, for variants within different domains of the protein. The authors should test/validate their prediction using mutants that are not included in the training sets.

In general, I think that the computational experiments and analysis described are developed sufficiently within the manuscript to inspire confidence that the findings produce meaningful results, and as mentioned above, this paper describes an interesting concept but the fact that this analysis was done for only a single type of protein and that training and validation were performed on the same variants are a major issues and must be addressed before acceptance of this paper for publication.

**********

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: 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 Mr. Phul,

We are pleased to inform you that your manuscript 'Predicting the Functional Impact of KCNQ1 Variants with Artificial Neural Networks' 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,

Joanna Slusky, Ph.D.

Associate Editor

PLOS Computational Biology

Arne Elofsson

Deputy 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: I thank the authors for their detailed responses to my comments and the updates to the paper. I already liked the original version and find the revised version strengthened and more clear. Congratulations on a nice piece of work.

Reviewer #2: Accept.

**********

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: No