QSAR models for antioxidant activity of new coumarin derivatives

^$Presented at the 8^th International Symposium on Computational Methods in Toxicology and Pharmacology Integrating Internet Resources, CMTPI-2015, 21–25 June, 2015, Chios, Greece.

Abstract

This study presents 37 new antioxidant coumarin derivatives and strategies for structural modification to improve their antioxidant activities, the main ferric-reducing antioxidant power (FRAP) assay used to evaluate their antioxidant properties and the generation of validated quantitative structure–activity (antioxidant activity) relationship (QSAR) models. In an attempt to generate QSAR models, structures of all coumarin derivatives in the data set were fully optimized by semi-empirical PM6 method using SPARTAN 10 software. Descriptors were calculated by DRAGON 6.0 software. Multiple linear regression (MLR) models were developed with different training/test set combinations using QSARINS 2.2.1 software. Robustness, reliability and predictive power of the models were tested by internal and external validations. Applicability domain of the best two-descriptor model (n_TR = 30; r² = 0.924; RMSE_TR = 0.213; n_TEST = 7; r²_ext = 0.887; RMSE_ext = 0.255; CCC_ext = 0.939) was determined. Descriptors appeared in the model revealed that complexity, H-bond donor and lipophilic character are important parameters in describing the antioxidant activity. Apart from the compounds in the data set, we also designed 31 new antioxidant coumarin derivatives and predicted their antioxidant activity using the best two-descriptor model. Most of these compounds are promising antioxidants

Keywords:

• QSAR
• in silico modelling
• multiple linear regression
• coumarin derivatives
• FRAP assay

Acknowledgements

S.S. Erdem acknowledges Marmara University, Scientific Research Project Committee for the support of the project FEN-C-YLP-130515-0186. The support of this study by Boğaziçi University Research Fund (No: 8502) is also acknowledged. The authors would like to thank Prof. Gramatica and Dr. Chirico for providing the QSARINS (version 2.2.1) software. P. Erzincan would like to thank TÜBİTAK-BİDEB 2210-C.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

^$ Presented at the 8^th International Symposium on Computational Methods in Toxicology and Pharmacology Integrating Internet Resources, CMTPI-2015, 21–25 June, 2015, Chios, Greece.