Zahra Garkani-Nejad; Marziyeh Poshteh-Shirani
Volume 3, Issue 2 , September 2016, , Pages 116-126
Abstract
A 2D image approach has been used to predict 13C NMR chemical shifts of β-naphthalene derivatives. In multivariate image analysis-Quantitative structure property relationship (MIA-QSPR) study, descriptors correlating with dependent variable are pixels (binaries) of 2D chemical structures; Variant ...
Read More
A 2D image approach has been used to predict 13C NMR chemical shifts of β-naphthalene derivatives. In multivariate image analysis-Quantitative structure property relationship (MIA-QSPR) study, descriptors correlating with dependent variable are pixels (binaries) of 2D chemical structures; Variant pixels in the structures (substitutes) account to explained variance in the property (chemical shifts). A case study is carried out in order to predict 13C NMR chemical shifts of 10 carbon positions of 24 mono substituted β-naphthalenes. The resulted descriptors were subjected to principal component analysis (PCA) and the most significant principal components (PCs) were extracted. Then, MIA-QSPR modeling was done by means of principal component regression (PCR) and principal component –artificial neural network (PC-ANN) methods. A correlation ranking procedure is proposed here to select the most relevant set of PCs as inputs for PCR and PC-ANN modeling methods. Here, the 13C chemical shifts of studied compounds were predicted using density functional theory (DFT) calculations, too. The widely applied method of gauge included atomic orbital (GIAO) B3LYP/6-311++ G have been used. The performance of the GIAO was also compared with PCR and PC-ANN models. Results showed the superiority of the PC-ANN over GIAO and PCR models. Finally, 13C NMR chemical shifts of studied compounds were calculated using ChemDraw program.
