Prediction of Sensor Ability Based on Chemical Formula: Possible Approaches and Pitfalls
, George A. Gamov 1,*, Michail M. Lukanov 2, Alexander A. Ksenofontov 2, Natalia A. Bumagina 2 and Elena V. Antina 2Round 1
Reviewer 1 Report
Overall, the object and scope of the formula and the content of the listed pictures have not been clearly described. Also, the author has described the software in less detail. Other detailed comments: 1. Introduction: The vertical coordinates of Figure 1 should be used in the vertical version and placed on the leftmost side. 2. Abstract: The word "different approaches" in the first sentence of the abstract is not appropriate and should be changed to "different algorithms", because the conclusion shows that four different algorithms are studied in one software. 3.Results and Discussion : Inconsistent punctuation after formula.Author Response
First, we would like to thank the Reviewer for the attention paid to our contribution, evaluation and comments made to help us in improving our manuscript. Please, find the detailed answers on your comments below:
Q.: “Overall, the object and scope of the formula and the content of the listed pictures have not been clearly described.”
A.: The captions to Figure 3 was expanded to make its content more clear. Moreover, the missing description to Eq. (6) (Euclidean distance) was also added.
Q.: “Also, the author has described the software in less detail.“
A.: The code developed is freely available online: https://gitlab.com/GeorgeGamov/prediction-of-sensing-ability. We guarantee the possibility to further use it by any person and help and consultations if somebody would wish to use it directly instead of applying the Telegram bot (which is also freely available).
Q.: “Introduction: The vertical coordinates of Figure 1 should be used in the vertical version and placed on the leftmost side”
A.: The indicated problem is corrected in the revised manuscript.
Q.: “Abstract: The word "different approaches" in the first sentence of the abstract is not appropriate and should be changed to "different algorithms"
A.: The indicated problem is corrected in the revised manuscript.
Q.: “Results and Discussion : Inconsistent punctuation after formula.”
A.: The punctuation after formulae was checked and corrected.
The changes made in accordance with your commentary are highlighted in the text with yellow.
Reviewer 2 Report
Comments:
1. Few more additional lines should be added to make the concept of “SMILES” more
understanding.
2. Under the fingerprint similarity methods, the concept of Sorensen–Dice coefficient overlap coefficient should also be explained along with a brief comparison with the Euclidian distance method.
3. Caption for figure-2 should be a bit elaborative, explaining the process.
4. Reference for figure-3 should be added.
Author Response
First, we would like to thank the Reviewer for the attention paid to our contribution, evaluation and comments made to help us in improving our manuscript. Please, find the detailed answers on your comments below:
Q.: “Few more additional lines should be added to make the concept of “SMILES” more understanding.”
A.: Lines 95-108 were added in the revised manuscript to clarify the concept of SMILES.
Q.: “Under the fingerprint similarity methods, the concept of Sorensen–Dice coefficient overlap coefficient should also be explained along with a brief comparison with the Euclidian distance method.”
A.: The explanation of Sorensen–Dice coefficient and overlap coefficient is provided (lines 207-214). The main difference between Euclidian distance method and the overlap coefficient method or Sorensen–Dice coefficient method is that the former is based on the determination of distance between two point in some conditional space, while the latter considers the elements in common of two different sets. These two methods are more close to the Tanimoto coefficient.
We plan in future development of our software, first, to expand the database with new entries; second, to introduce new methods for similarity search, or, if the database size would be sufficient, to return to the correlation search between molar and ionic descriptors.
Q.: “Caption for figure-2 should be a bit elaborative, explaining the process.”
A.: The caption for Fig. 2 was expanded
Q.: “Reference for figure-3 should be added.”
A.: The indicated problem is corrected in the revised manuscript (line 296).
The changes made in accordance with your commentary are highlighted in the text with green.
Reviewer 3 Report
This is a very unique and nice manuscript. The author managed a library of around 965 chemical sensors and developed a Telegram bot out of it, which can predict the sensing ability of the chemical compounds for metal ions. This manuscript can be very helpful for researchers planning to design a metal ions sensor. I just wonder about data privacy i.e. the chemical structure fed to the Telegram bot by some researchers. If that can be handled securely then it would be really good.
Author Response
We would like to thank the Reviewer for the attention paid to our contribution, and its evaluation. We appreciate the kind words of support. Regarding the question about data privacy, we would like to assure that nobody from the Authors has an access to the data input by other users. In addition, such a large and well-known for its privacy-protection measures company as Telegram, in our opinion, can provide the security level, which is unachievable to us, had we decide to create a separate site for the aim of hosting GUI for our software.
Round 2
Reviewer 1 Report
I have read authors' response online. They have answered all my questions, and those answers were clear and proper. I further checked the revised manuscript, and found that Figure 1 & 2 have been carefully reorganized and revised. Also, equations 5 and 8 have been corrected. The paragraph in line 95 have been modified. Based on this, I suggest that this paper could be accepted at current status.
