4D-QSAR analysis and pharmacophore modeling for alkynylphenoxyacetic acids as CRTh2 (DP2) receptor antagonists


Köprü S., Sarıpınar E.

TURKISH JOURNAL OF CHEMISTRY, vol.42, pp.1577-1597, 2018 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 42
  • Publication Date: 2018
  • Doi Number: 10.3906/kim-1801-86
  • Journal Name: TURKISH JOURNAL OF CHEMISTRY
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, TR DİZİN (ULAKBİM)
  • Page Numbers: pp.1577-1597
  • Keywords: Electron conformational genetic algorithm, 4D-QSAR, CRTh2 receptor antagonist, pharmacophore, drug design, alkynylphenoxyacetic acid, GENETIC ALGORITHM, BIOACTIVITY PREDICTION, QSAR MODELS, IDENTIFICATION, DERIVATIVES, VALIDATION, SELECTION, POTENT, PROSTAGLANDIN-D2, OPTIMIZATION
  • Erciyes University Affiliated: Yes

Abstract

In this study, we performed the pharmacophore modeling and 4D-QSAR research of alkynylphenoxyacetic acid analogues as CRTh2 receptor opponent agents by utilizing the electron conformational genetic algorithm method. Quantum chemical calculations and conformational analyses of the compounds were carried out at HF/6-31G* level. Then electron conformational matrices of congruity were prepared for each conformer of each compound, which are represented by electronic and structural properties. As a result of the comparison of the matrices that are called electron conformational submatrices of activity, the pharmacophoric group of the compounds responsible for the activity was found at the determined tolerance intervals. The genetic algorithm and nonlinear least squares regression methods were applied to estimate the conjectural activity and investigate the most reliable molecular identifiers as feature selection from a large parameter pool. The compounds in the dataset were randomly segregated for training (61 compounds) and test sets (25 compounds) for statistical analysis. Validation of the 4D-QSAR model was appraised by the leave-one-out cross-validation technique. For the best model the r(training)(2), r(test)(2), q(2), q(ext1)(2), q(ext2)(2), and q(ext3)(2) values were found to be 0.8580, 0.8571, 0.8105, 0.8282, 0.8145, and 0.8475, respectively.