Next Article in Journal
An Eco-Driving Algorithm for Interoperable Automatic Train Operation
Previous Article in Journal
Charge Transport Behavior of Al-Doped ZnO Incorporated with Reduced Graphene Oxide Nanocomposite Thin Film
 
 
Article
Peer-Review Record

Tree2C: A Flexible Tool for Enabling Model Deployment with Special Focus on Cheminformatics Applications

Appl. Sci. 2020, 10(21), 7704; https://doi.org/10.3390/app10217704
by Alessandro Pedretti *, Angelica Mazzolari, Silvia Gervasoni and Giulio Vistoli
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Appl. Sci. 2020, 10(21), 7704; https://doi.org/10.3390/app10217704
Submission received: 6 October 2020 / Revised: 28 October 2020 / Accepted: 29 October 2020 / Published: 30 October 2020
(This article belongs to the Section Chemical and Molecular Sciences)

Round 1

Reviewer 1 Report

The manuscript is well done, bus some minor corrections are needed:

Introduction: is necessary that authors to emphasise (with recent references) the importance/critical role of AI in:

  • drug design/development
  • the importance of BBB penetration
  • health risks concerns (with accent on mutagenicity)

---------------------------------

Line 38: “Python[4]” – a space is missing here

Line 41: explain “HTS” (even if it appears only here!)

Line 43: explain “ADME” (even if it appears only here!)

Line 45: “However and similarly” – sounds weird... can you reformulate?

Line 51: “problematic step. [9]” – put the reference before the “.”

Line 62: reference is missing for Weka program (is the version relevant?... if YES = add-it)

Line 65, 72, 117, 119-121 (etc...): is “VEGA environment” same thing with “VEGA environment” and “VEGA software” – if YES, please use the same syntagm

Line 118-119, 148 (etc...): use for MOPAC version the syntagm used by its authors, reference is missing (use the instruction from official site to refer/cite: http://openmopac.net/)

Line 172: “non-permeants. [11]” – put the reference before the “.”

Line 187: “...thus markedly reducing the computational costs” – questions: there is any negative effect (e.g.: scoring, etc...)?

Author Response

Dear Reviewer,

We thank you for the valuable suggestions. Here is the description of the amendments made in the revised version according your requests. All modifications are highlighted in yellow to track changes.

 

The manuscript is well done, bus some minor corrections are needed:

Introduction: is necessary that authors to emphasise (with recent references) the importance/critical role of AI in:

  • drug design/development
  • the importance of BBB penetration
  • health risks concerns (with accent on mutagenicity)

According these suggestions, we improved the introduction adding the following sentences:

“Along with the availability of suitable databases and reference studies, these two examples were chosen by considering the remarkable medicinal interest in predicting both the capacity of a given molecule to penetrate to CNS as well as its potential toxic concerns as exemplified by mutagenicity. In more detail, the capacity to predict the BBB permeation based on the physicochemical properties of a given permeant is of crucial relevance to assist the rational design of molecules developed for neurological disorders. To this end, different approaches have been applied including standard linear regressions [13], machine learning methods [14] or tailored algorithms (as seen with the recent BBB score [15]). Notably, some of the developed predictive models are freely available as on line services [16].

Similarly, the prediction of potential toxicity concerns of novel molecules is of crucial relevance to minimize the health risks during drug development [17] as well as to assure the overall safety of the industrial chemicals [18]. In detail, the in silico mutagenicity prediction is gaining a key role in the last years since this can be also used for regulatory purposes provided that the developed models assure appropriate robustness and predictive power with a well-defined applicability domain and considered endpoint [19]. Hence, the remarkable number of predictive studies recently published comes as no surprise: the applied computational techniques include classic QSAR methods, rule-based methods, AI algorithms and consensus approaches [20].”

---------------------------------

Line 38: “Python[4]” – a space is missing here

Line 41: explain “HTS” (even if it appears only here!)

Line 43: explain “ADME” (even if it appears only here!)

Line 45: “However and similarly” – sounds weird... can you reformulate?

Line 51: “problematic step. [9]” – put the reference before the “.”

We updated the manuscript according all these suggestions and we explained the meaning of both HTS and ADME acronyms.

Line 62: reference is missing for Weka program (is the version relevant?... if YES = add-it)

We added the reference for Weka [3].

Line 65, 72, 117, 119-121 (etc...): is “VEGA environment” same thing with “VEGA environment” and “VEGA software” – if YES, please use the same syntagm

We changed “VEGA software” to “VEGA environment”.

Line 118-119, 148 (etc...): use for MOPAC version the syntagm used by its authors, reference is missing (use the instruction from official site to refer/cite: http://openmopac.net/)

We updated the paper and added the reference.

Line 172: “non-permeants. [11]” – put the reference before the “.”

We updated the paper according this suggestion.

Line 187: “...thus markedly reducing the computational costs” – questions: there is any negative effect (e.g.: scoring, etc...)?

We not found any significant effect on the overall predictive power. Therefore, we modified the sentence: “...thus markedly reducing the computational costs without affecting the overall predictive power”.

 

We do believe that the revised version fully answers your comments.

Reviewer 2 Report

Tree2C addresses an important problem of deploying AI based predictive

models that can be utilized for making new predictions, especially in

the field of medicinal chemistry. Of course, machine learning models

can only generate data of value when the insights of the models are

delivered to end users. Tree2C automatically translates tree-based

predictive models generated by the Weka software into a source code

which can be easily used by end users to generate standalone

applications. The source code generated by Tree2C is formatted for

several commonly used programming languages including C++, Fortran,

Java, Python, PHP and others, which is indeed

commendable. Functionality of the software has been clearly

demonstrated by the use of ML model generated using random forest

algorithm for prediction of permeability and mutagenicity. Authors

have also shown the ability to use the MOPAC program through scripts

generated by Tree2C which is useful to evaluate important descriptors

of a chemical dataset. The details of the program have been neatly

described in the article and the software under consideration will

definitely prove useful in various aspects of drug design in

pharmaceutical sector.

 

 

Author Response

We thank you for the time you spent in reviewing our paper. 

Back to TopTop