Identifying Potent Fat Mass and Obesity-Associated Protein Inhibitors Using Deep Learning-Based Hybrid Procedures
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsDear Author,
After reviewing your paper, I found it interesting and I think it can be an appropriate addition for the journal. However, minor to moderate revisions are recommended. Please see below for details:
Summary:
- The study focuses on the Fat Mass and Obesity related protein (FTO), which is known for catalyzing metal-dependent modifications of nucleic acids, particularly the demethylation of methyl adenosine in mRNA molecules. FTO is identified as a potential target for developing anticancer therapies.
- The main objective of the research was to identify small molecular chemical compounds that can inhibit the FTO protein. The team utilized crystal structures of FTO complexed with ligands and employed a combination of deep learning (using DeepBindGCN) and autodock vina on the ZINC database to identify potential inhibitors. Three compounds were identified as potential FTO protein inhibitors: ZINC000003643476, ZINC000000517415, and ZINC000001562130.
- These compounds underwent comprehensive analysis, including 100 nanoseconds of molecular dynamics and binding free energy calculations. The study's results indicated that these three compounds could effectively target the human fat mass and obesity protein, suggesting a potential pathway for exploring other chemicals that interact with FTO. The research provides a foundation for biochemical studies to evaluate the effectiveness of these compounds, which could contribute significantly to the development of new obesity treatment strategies.
- The paper highlights the use of advanced techniques like deep learning and molecular dynamics in drug discovery, emphasizing the evolving landscape of biomedical research where computational methods are increasingly playing a critical role in identifying and validating new therapeutic compounds.
Strengths:
- Sound research methodology
- Relevant topic
- Interesting and significant subject
- Appropriate presentation of results
Weaknesses:
- Lack of hypothesis and/or research questions
- Lack of in-depth discussion
- Significance and implications are not clearly noted
Comments and suggestions:
Comment 1: Please note the knowledge gap in this domain and how your study contributes to fill this gap
Comment 2: Consider adding research questions that are answered by your study.
Comment 3: Please expand the discussion section with additional references (published 2021-2023), note the social, practical and other implications of your study.
Comment 4: Please discuss the limitations and advantages of your study.
Comment 5: Consider noting guidelines and suggestions for future research in this domain.
Comment 6: Please expand the Introduction section with additional references that provide a solid theoretical background.
Kind regards,
Reviewer
Author Response
Comment
Please note the knowledge gap in this domain and how your study contributes to fill this gap.
Response
We thank the referee for useful comment. We have added a paragraph that provide the knowledge gap in this domain and how our study contributes to fill this gap as follows: “Identifying a suitable ligand-targeting FTO protein is crucial for the development of chemo-therapeutic medicines to combat obesity and cancer. Scientists worldwide employ many methodologies to discover a potent inhibitor for the FTO protein. In this study, we utilize a combination of deep learning-based methods and classic molecular docking techniques to investigate the FTO protein as a target. Our strategy involves systematically screening a database of small chemical compounds.”
Comment
Consider adding research questions that are answered by your study.
Response
We thank the referee for useful comment. We have added a paragraph to address this as follows: “Identifying a suitable ligand-targeting FTO protein is crucial for the development of chemo-therapeutic medicines to combat obesity and cancer. Scientists worldwide employ many methodologies to discover a potent inhibitor for the FTO protein. The findings of our study indicate the identification of three candidate inhibitors that might effectively target the human fat mass and obesity protein. The results of this study have the potential to facilitate the exploration of other chemicals that can interact with FTO.”
Comment
Please expand the discussion section with additional references (published 2021-2023), note the social, practical and other implications of your study.
Response
As per referee suggestion, discussion section has been revised suitably.
Comment
Please discuss the limitations and advantages of your study.
Response
As per referee suggestion, limitations and advantages of our study is added in the discussion section as follows: “Nevertheless, contemporary deep learning algorithms have already been utilized to identify inhibitors and exhibit immense potential in the field of drug development. Nevertheless, there are still other constraints that require consideration, encompassing both efficiency and accuracy.
The increasing availability of deep learning algorithms and the growing collection of experimental protein-ligand interaction data will significantly enhance the use of deep learning for virtual drug screening.”
Comment
Consider noting guidelines and suggestions for future research in this domain.
Response
We have added guidelines and suggestions for future research in the abstract as follows: “The results of this study have the potential to facilitate the exploration of other chemicals that can interact with FTO. Conducting biochemical studies to evaluate the effectiveness of these compounds may contribute to the improvement of fat mass and obesity treatment strategies.”
Comment
Please expand the Introduction section with additional references that provide a solid theoretical background.
Response
The introduction section has been revised carefully by adding more details on computer aided drug discovery and hybrid approaches.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe manuscript presents a methodical approach to identifying FTO inhibitors via crystallographic data and a combination of deep learning and docking methods, yielding three potential small molecule inhibitors from the ZINC database. These were further evaluated through molecular dynamics simulations and binding free energy calculations, providing valuable insights into their potential efficacy in modulating FTO activity for obesity treatment strategies.
However, it is imperative to address the broader physiological implications of FTO inhibition due to its vital role in various biological processes and its link to multiple human pathologies. The possibility of off-target effects or systemic toxicity warrants a thorough discussion, with an emphasis on a risk-benefit analysis in the context of therapeutic use.
Furthermore, to contextualize the significance of the findings, a comparative analysis with existing FTO inhibitors would be highly beneficial. Such a comparison would enhance the readers' understanding of the novel inhibitors' relative efficacy and could guide future research directions. Incorporating these elements into the discussion would greatly strengthen the manuscript's contribution to the field.
Author Response
Comment
The manuscript presents a methodical approach to identifying FTO inhibitors via crystallographic data and a combination of deep learning and docking methods, yielding three potential small molecule inhibitors from the ZINC database. These were further evaluated through molecular dynamics simulations and binding free energy calculations, providing valuable insights into their potential efficacy in modulating FTO activity for obesity treatment strategies.
However, it is imperative to address the broader physiological implications of FTO inhibition due to its vital role in various biological processes and its link to multiple human pathologies. The possibility of off-target effects or systemic toxicity warrants a thorough discussion, with an emphasis on a risk-benefit analysis in the context of therapeutic use.
Furthermore, to contextualize the significance of the findings, a comparative analysis with existing FTO inhibitors would be highly beneficial. Such a comparison would enhance the readers' understanding of the novel inhibitors' relative efficacy and could guide future research directions. Incorporating these elements into the discussion would greatly strengthen the manuscript's contribution to the field.
Response
We thank the referee for his valid comments. We have revised introduction and discussion section thoroughly to answer the key queries raised.
Reviewer 3 Report
Comments and Suggestions for AuthorsMore studies should be examined under the title of literature summary. The reviewed studies should also be presented with a table.
The manuscript needs extensive revision for related work.
Introduction should clearly state the proposal and major contributions.
The novelty of the work proposed in the paper needs to be discussed.
In result analysis, the author must explain the results in detail, with observations and general reasons. It is beneficial that the author includes the scientific basis for each finding.
Comments on the Quality of English LanguagePlease make sure your paper has necessary language proof-reading.
Conclusion should not mere summarize the paper
Author Response
Comment
More studies should be examined under the title of literature summary. The reviewed studies should also be presented with a table.
Response
We thank the referee for his valid comments. We have revised introduction and discussion thoroughly and added related key references.
Comment
The manuscript needs extensive revision for related work.
Response
The manuscript revised extensively for citing related work.
Comment
Introduction should clearly state the proposal and major contributions.
Response
As per referee suggestion, we have clearly stated the proposal and major contributions as follows:
“We have introduced an interesting approach for drug discovery called hybrid drug virtual screening. This technology combines deep machine learning, molecular docking, and molecular dynamics simulation to identify potential therapeutic candidates from chemical databases. This computational analysis examines candidate compounds with the potential to inhibit the FTO protein. The prediction is improved in terms of accuracy and efficiency by combining three computing approaches: (i) virtual screening of the existing chemicals database, (ii) deep learning, and (iii) molecular dynamics and free energy calculations. In this study, we aim to discover small molecular chemical compounds that exhibit inhibitory properties against FTO. The computational analysis of ligands complexed with human FTO demonstrates that the newly discovered small molecule exhibits a specific binding mode with FTO. The discovery of the small molecule presents potential avenues for advancing the development of FTO inhibitors that are more discerning and efficacious.”
Comment
The novelty of the work proposed in the paper needs to be discussed.
Response
The novelty of proposed work is discussed now as follows:
“We have employed a novel approach for drug discovery called hybrid drug virtual screening. This technology combines deep machine learning, molecular docking, and molecular dynamics simulation to identify potential therapeutic candidates from chemical databases. This computational analysis examines candidate compounds with the potential to inhibit the FTO protein. The prediction is improved in terms of accuracy and efficiency by combining three computing approaches: (i) virtual screening of the existing chemicals database, (ii) deep learning, and (iii) molecular dynamics and free energy calculations. “
Comment
In result analysis, the author must explain the results in detail, with observations and general reasons. It is beneficial that the author includes the scientific basis for each finding.
Response
As per referee suggestion, we explain the results in detail and included the scientific basis for each finding as follows:
“The hydrophobic interactions play a crucial role in ligand-protein interactions [61]. The hydrophobic carbon composition of a ligand mostly determines whether the ligand can initially enter the active site [62]. The precise conformation of these hydrophobic carbons with the geometry of the active site also guarantees the absence of undesired lig-and-protein interactions resembling the target [63].
Charged residues were discovered to enhance the high affinity binding. Additionally, they play a crucial role in "electrostatic steering," a technique that allows electrostatic forces to guide a ligand protein towards a binding site on the receptor protein, resulting in a signif-icant increase in the rate of association [64,65].”
Comment
Please make sure your paper has necessary language proof-reading.
Response
The manuscript has been proof read by a native English speaker for errors.
Comment
Conclusion should not mere summarize the paper
Response
The conclusion section has been revised appropriately.
Reviewer 4 Report
Comments and Suggestions for AuthorsDeepBindGCN, a deep learning model based on Graph Convolution Networks need to discuss and explore more.
How to evaluate your proposed method? What is the standard value in Table 1 and what is the meaning?
This research need to rewrite again very carefully, and describes what is the major novelty in this research.
Need to add more discussions, not only one paragraph.
Explore more about the deep learning-based hybrid procedures.
Add the formula or equations.
Author Response
Comment
DeepBindGCN, a deep learning model based on Graph Convolution Networks need to discuss and explore more.
Response
As per referee suggestion, we have added more details about DeepBindGCN model as follows:
“Moreover, the GCN has proven to be effective in predicting protein-ligand interactions. In our recent work, we suggested a screening process that combines DeepBindGCN with other techniques to identify molecules with a high affinity for binding, using TIPE3 and PD-L1 dimer as examples to demonstrate this approach. One advantage of DeepBindGCN is its ability to operate without relying on specific docking conformations. Additionally, it effectively preserves both spatial information and physical-chemical characteristics.In this study, we utilized the pocket residues or ligand atoms as the nodes and established edges based on the nearby information to create a comprehensive representation of the protein pocket or ligand information. Furthermore, the model utilizing pre-trained molecular vec-tors exhibited superior performance compared to the model using one-hot representation. The method demonstrated a similar level of predictive ability as the most advanced affin-ity prediction models that depend on the three-dimensional complex. DeepBindGCN is a robust technique for predicting protein-ligand interactions and can be applied in several significant large-scale virtual screening scenarios.”
Comment
How to evaluate your proposed method? What is the standard value in Table 1 and what is the meaning?
Response
In our recent work, we suggested a screening process that combines DeepBindGCN with other techniques to identify molecules with a high affinity for binding, using TIPE3 and PD-L1 dimer as examples to demonstrate this approach (https://doi.org/10.3390/molecules28124691).
Table 1 displays the DeepBindGCN_BC, which is a binary classifier that distinguishes between binders and non-binders, as well as the DeepBindGCN_RG, which is a process that predicts the affinity between proteins and ligands. Throughout the training and application process, we saw that the input preparation and model architecture remained very constant. The only difference was that one output was a binary value ranging from 0 to 1, indicating categorization of binding and non, while the other output was a continuous value representing affinity prediction.
Comment
This research need to rewrite again very carefully, and describes what is the major novelty in this research.
Response
The novelty in the research is stated clearly in the revised manuscript.
“We have employed a novel approach for drug discovery called hybrid drug virtual screening. This technology combines deep machine learning, molecular docking, and molecular dynamics simulation to identify potential therapeutic candidates from chemical databases. This computational analysis examines candidate compounds with the potential to inhibit the FTO protein. The prediction is improved in terms of accuracy and efficiency by combining three computing approaches: (i) virtual screening of the existing chemicals database, (ii) deep learning, and (iii) molecular dynamics and free energy calculations. “
Comment
Need to add more discussions, not only one paragraph.
Response
We thank the referee for his valid comments. We have revised introduction and discussion thoroughly and added related key references.
Comment
Explore more about the deep learning-based hybrid procedures. Add the formula or equations.
Response
More details about deep learning-based hybrid procedures is added in the revised manuscript.
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsDear Authors,
Good job on the revision. The paper is now well-established with a sound methodology, high readability, and detailed explanation of the research. I don't have further comments.
Kind regards,
Reviewer
Author Response
Comment
Good job on the revision. The paper is now well-established with a sound methodology, high readability, and detailed explanation of the research. I don't have further comments.
Response
We thank the referee for his valuable time to evaluate our research.
Reviewer 2 Report
Comments and Suggestions for AuthorsRevised based on given comments.
Author Response
Comment
Revised based on given comments.
Response
We thank the referee for his valuable time to evaluate our research.
Reviewer 3 Report
Comments and Suggestions for Authors.Authors are suggested to use high resolution vector for each figures.
There should be links between paragraphs and sections which should lead to the next one.
Author Response
Comment
Authors are suggested to use high resolution vector for each figure.
Response
We thank the referee for his valid comments. We have attached high resolution figures as separate files.
Comment
There should be links between paragraphs and sections which should lead to the next one.
Response
The connecting statements have been added suitably in the revised manuscript.
Reviewer 4 Report
Comments and Suggestions for Authors-Pleased to discuss the major contribution of the research
-Explore the other deep learning method
- add a related reference:
https://doi.org/10.3390/bdcc6040106
DOI: 10.1038/s41598-023-35431-x
10.1109/ACCESS.2023.3309410
https://doi.org/10.1016/j.compag.2023.108481
DOI: 10.1109/TIA.2021.3126272
DOI: 10.1109/TNNLS.2020.3046629
Author Response
Comment
-Pleased to discuss the major contribution of the research
-Explore the other deep learning method
- add a related reference:
https://doi.org/10.3390/bdcc6040106
DOI: 10.1038/s41598-023-35431-x
10.1109/ACCESS.2023.3309410
https://doi.org/10.1016/j.compag.2023.108481
DOI: 10.1109/TIA.2021.3126272
DOI: 10.1109/TNNLS.2020.3046629.
Response
As per referee suggestion, the related deep learning methods were added in the revised manuscript suitably (Ref: 49, 50, 52, 53, 54 and 55).
Round 3
Reviewer 4 Report
Comments and Suggestions for AuthorsThis paper can be accepted now. The Author already revised the paper based on the reviewer's comment.