3D-QSAR studies of some reversible Acetyl cholinesterase inhibitors based on CoMFA and ligand protein interaction fingerprints using PC-LS-SVM and PLS-LS-SVM. (August 2017)

• Record Type: Journal Article
• Authors: Ghafouri, Hamidreza; Ranjbar, Mohsen; Sakhteman, Amirhossein
• Is Part Of: Computational biology and chemistry. Volume 69(2017)
• Page Start: 19
• Page End: 27
• Publication Date: 2017-08
• Subjects: Acetylcholinesterase -- Field descriptors -- Interaction descriptors; Chemistry -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; Biochemistry -- Data processing; Biology -- Data processing; Molecular biology -- Data processing; Periodicals; Electronic journals; 542.85
• DOI: 10.1016/j.compbiolchem.2017.05.001
• Languages: English
• ISSNs: 1476-9271

Abstract

Graphical abstract: Highlights: Activity of Some acetylcholinesterase inhibitors were studies in terms of electrostatic, steric and interaction features. Field descriptors approach in 3D-QSAR was compared to that of interaction descriptors. The more robustness and predictive ability of field descriptors approach was verified. Based on external validation criteria, both methods can generate reasonable QSAR models. Using interaction descriptor approach can be implemented in novel 3D-QSAR applications. Abstract: A great challenge in medicinal chemistry is to develop different methods for structural design based on the pattern of the previously synthesized compounds. In this study two different QSAR methods were established and compared for a series of piperidine acetylcholinesterase inhibitors. In one novel approach, PC-LS-SVM and PLS-LS-SVM was used for modeling 3D interaction descriptors, and in the other method the same nonlinear techniques were used to build QSAR equations based on field descriptors. Different validation methods were used to evaluate the models and the results revealed the more applicability and predictive ability of the model generated by field descriptors (Q 2 LOO-CV = 1, R 2 ext = 0.97). External validation criteria revealed that both methods can be used in generating reasonable QSAR models. It was concluded that due to ability of interaction descriptors in prediction of binding mode, using this approach can be implemented in future 3D-QSAR softwares.