Despite the increasing role played by artificial intelligence methods (AI) in pharmaceutical sciences, model deployment remains an issue, which only can be addressed with great difficulty. This leads to a marked discrepancy between the number of published predictive studies based on AI methods and the models, which can be used for new predictions by everyone. On these grounds, the present paper describes the Tree2C tool which automatically translates a tree-based predictive model into a source code with a view to easily generating applications which can run as a standalone software or can be inserted into an online web service. Moreover, the Tree2C tool is implemented within the VEGA environment and the generated program can include the source code to calculate the required attributes/descriptors. Tree2C supports various programming languages (i.e., C/C++, Fortran 90, Java, JavaScript, JScript, Lua, PHP, Python, REBOL and VBScript and C-Script). Along with a detailed description of the major features of this tool, the paper also describes two examples which are aimed to predict the blood–brain barrier (BBB) permeation as well as the mutagenicity. They permit a clear evaluation of the potentials of Tree2C and of its related features as implemented by the VEGA suite of programs. The Tree2C tool is available for free.

Tree2c: A flexible tool for enabling model deployment with special focus on cheminformatics applications / A. Pedretti, A. Mazzolari, S. Gervasoni, G. Vistoli. - In: APPLIED SCIENCES. - ISSN 2076-3417. - 10:21(2020 Nov 01), pp. 7704.1-7704.8. [10.3390/app10217704]

Tree2c: A flexible tool for enabling model deployment with special focus on cheminformatics applications

A. Pedretti
Primo
Conceptualization
;
A. Mazzolari
Secondo
Investigation
;
S. Gervasoni
Penultimo
Investigation
;
G. Vistoli
Ultimo
Supervision
2020-11-01

Abstract

Despite the increasing role played by artificial intelligence methods (AI) in pharmaceutical sciences, model deployment remains an issue, which only can be addressed with great difficulty. This leads to a marked discrepancy between the number of published predictive studies based on AI methods and the models, which can be used for new predictions by everyone. On these grounds, the present paper describes the Tree2C tool which automatically translates a tree-based predictive model into a source code with a view to easily generating applications which can run as a standalone software or can be inserted into an online web service. Moreover, the Tree2C tool is implemented within the VEGA environment and the generated program can include the source code to calculate the required attributes/descriptors. Tree2C supports various programming languages (i.e., C/C++, Fortran 90, Java, JavaScript, JScript, Lua, PHP, Python, REBOL and VBScript and C-Script). Along with a detailed description of the major features of this tool, the paper also describes two examples which are aimed to predict the blood–brain barrier (BBB) permeation as well as the mutagenicity. They permit a clear evaluation of the potentials of Tree2C and of its related features as implemented by the VEGA suite of programs. The Tree2C tool is available for free.
AI methods; BBB prediction; Classification algorithms; Model deployment; Mutagenicity; Tree-based methods
Settore CHIM/08 - Chimica Farmaceutica
Article (author)
File in questo prodotto:
File Dimensione Formato  
Appl.Sci._10_2020.pdf

accesso aperto

Tipologia: Publisher's version/PDF
Dimensione 8.64 MB
Formato Adobe PDF
8.64 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

Caricamento pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/2434/789603
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 1
  • ???jsp.display-item.citation.isi??? 1
social impact