QSAR study of catechol-based derivatives for urease inhibitors

Fei Zhu, Hongzong Si, Peijian Zhang, Honglin Zhai

Abstract: To study the catechol derivatives that showed better inhibitory activity and lower toxicity for urease, quantitative structure activity relationship method was applied to study the relationship between molecular structure and inhibitory activity of catechol derivatives. More than forty molecular structures of catechol derivatives were obtained from recent studies. The data set was divided into training set and test set randomly. The training set was used to build models. Heuristic method in CODESSA software was applied to select appropriate molecular descriptors from the descriptors' pool. Next, a nonlinear regression model based on the descriptors and inhibitory activity (IC50) of catechol derivatives and a classification model to judge whether the catechol derivatives were active or not were built by using a novel machine learning technique gene expression programming. After that, these two models were checked and validated the generalization ability by using the test set. The classification accuracy and sensitivity and specificity of training set and test set of nonlinear classification model were all 100%. In nonlinear regression model, the correlation coefficient and mean squared error of training set were 0.84 and 0.11 and the correlation coefficient and mean squared error of test set were 0.81 and 0.49. The results showed that both models had good stability and predictive ability.

: Download (free)  CSE-PDF

Quantitative Structure Activity Relationship Study of Novel 1,2,3-triazole-derived Diarylpyrimidines of HIV-1 NNRTIs

Yu Gu, Hongzong Si, Peijian Zhang, Honglin Zhai

Abstract: The quantitative structure activity relationship approach was used to predict the activity of novel HIV-1 non-nucleoside reverse transcriptase inhibitors. These novel 1,2,3-triazole-derived diarylpyrimidines were reported recently. Heuristic method has been well implemented in CODESSA software to select appropriate descriptor subsets for quantitative structure activity relationship modeling. In addition, Gene expression programming approach was first employed to build nonlinear models with the descriptors. The test set had been used to verify the accuracy of the models. Finally the satisfied quantitative structure activity relationship models were found. These models will be useful in the future design and development of novel HIV-1 NNRTIs.

: Download (free)  CSE-PDF

Quantitative structure-activity relationship study of glutamyl cyclase inhibitors based on gene expression programming

Shuting Tian, Hongzong Si, Peijian Zhang, Honglin Zhai

Abstract: Glutamyl cyclase (QC) has been proposed as a potential therapeutic target for the treatment of Alzheimer's disease. In this study, gene expression programming (GEP) extended from genetic algorithms and genetic programming was employed to establish nonlinear quantitative structure activity relationship (QSAR) model with descriptors to predict the activity of 29 novel QC inhibitors. These descriptors were calculated in CODESSA software and selected from descriptors’ pool based on heuristic method. And five descriptors were chosen to establish multivariate linear regression model. Then a nonlinear QSAR model with correlation coefficient of 0.86 and 0.88 and mean error of 0.01 and 0.01 for the training set and test set was obtained by GEP. The results showed that the model established by GEP had better stability and predictive ability.

: Download (free)  CSE-PDF