Please login first
Machine Learning and Atom-Based Quadratic Indices for Proteasome Inhibition Prediction
* 1, 2, 3 , 4 , 2 , 3
1  Universidad Estatal Amazónica, Facultad de Ingeniería Ambiental, Paso lateral km 2 1/2 via Napo, Puyo, Ecuador
2  Unidad de Investigación de Diseño de Fármacos y Conectividad Molecular, Departamento de Química Física, Facultad de Farmacia, Universitat de València, Spain
3  Departament de Bioquímica i Biologia Molecular, Universitat de València, E-46100 Burjassot, Spain
4  School of Medicine and Pharmacy, Vietnam National University, Hanoi (VNU) 144 Xuan Thuy, Cau Giay, Hanoi, Vietnam


The atom-quadratic indices are used in this work together with some machine learning techniques that includes: support vector machine, artificial neural network, random forest and k-nearest neighbor. This methodology is used for the development of two quantitative structure-activity relationship (QSAR) studies for the prediction of proteasome inhibition. A first set consisting of active and non-active classes was predicted with model performances above 85% and 80% in training and validation series, respectively. These results provided new approaches on proteasome inhibitor identification encouraged by virtual screenings procedures.

Keywords: Atom-based quadratic index, classification and regression model, machine learning, proteasome inhibition, QSAR, TOMOCOMD-CARDD software
Comments on this paper
Humbert G. Díaz
comparative study?
Dear Profs. Casañola-Martin et al.

Thank you by your support to mol2net conference.
Do you carried out a comparative study of this problem with other indices of TOMO-COMD or DRAGON for instance?

Thank you very much
Gerardo Casañola-Martin
Dear Prof. Humbert González-Díaz

Thank you for your question. This work is preliminary, and other studies are in progress using TOMO-COMD. According to this, the same behavoir is observed, with the RF as the best techniques, for this type of dataset in the case of the TOMOCOMD indices. For DRAGON indices, a comparative should be done to corroborate if this behavoir remains for this dataset.

Marcus Scotti
Dear Prof Gerardo Maikel Casañola-Martin,

It is a nice study. Just a little suggestion: perhaps add some comparative study using several kinds of 3D and 2D descriptors as attributes, it could give some information, for your system, what kind of descriptor generates a better performance and maybe it is possible to analyse the reason.


Gerardo Casañola-Martin
Dear Prof. Marcus Scotti

Thanks a lot for your suggestion, I will take into consideration for comparative studies to obtain more information for the performance of the descriptors, and for the analysis of the reason according to the nature of the descriptors.

Best regards,