Drug design of new sigma-1 antagonists against neuropathic pain: A QSAR study using partial least squares and artificial neural networks

• Published in: Journal of molecular structure Vol. 1223; p. 129156
• Main Authors: Chiari, Laise P.A., da Silva, Aldineia P., de Oliveira, Aline A., Lipinski, Célio F., Honório, Kathia M., da Silva, Albérico B.F.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 05.01.2021
• Subjects: MLP-ANN; Neuropathic pain; PLS; QSAR; Sigma-1R; MLP-ANN; QSAR; PLS; Sigma-1R; Neuropathic pain
• ISSN: 0022-2860; 1872-8014
• DOI: 10.1016/j.molstruc.2020.129156

•Eight molecular and electronic descriptors were selected to construct QSAR models.•A sigma-1 receptor antagonist series is described by both PLS and ANN models.•New designed compounds exhibited relevant biological activity values in both models. Neuropathic pain is a cureless syndrome and affects considerably the life quality of people stricken by it. Drugs currently used for its treatment do not significantly reduce the symptoms and/or have many side effects. In the search for other therapeutic approaches, the sigma-1 receptor has been pointed out as a promising drug target for the treatment of neuropathic pain. As part of our effort to help the development of new therapeutic agents against neuropathic pain, we have applied techniques of quantitative structure-activity relationships (QSAR) to a series of compounds having the pyrimidine as scaffold using Partial Least Squares (PLS) and Artificial Neural Networks (ANN) to design new sigma-1R antagonists. Next, we have calculated a plethora of descriptors, which were selected from correlation matrix and genetic algorithm (AG). The selected descriptors were used to construct PLS and ANN models and, from them, various results were used to design new antagonists. At last, the designed compounds were subjected to the our QSAR models to predict their biological activity values. The new compounds exhibited significant biological affinity values, and among them we can highlight the compounds L2, L4, L14, L17, and L18 with excellent predicted pKi values confirmed by both PLS and ANN models. Therefore, the predictive ability of the PLS and ANN models here presented and their robustness allowed to extract important information that can be used in the design of new compounds as well as to predict their biological activity values. [Display omitted]