Quantitative Structure-Activity Relationship Model for Antioxidant Activity of Flavonoid Compounds in Traditional Chinese Herbs

Abstract

There are many diseases related to the excessive amount of free radical in human body produced by various metabolic functions. The generation of free radicals can be controlled by the presence of antioxidants. One of the largest phytochemical groups in herbs which commonly exhibit antioxidant activity is flavonoid. The structure of flavonoid compounds can be represented by various types of descriptors generated by the DRAGON software. A series of flavonoids with their antioxidant activities values in 2,2’-azinobis-3-ethylbenzothiazoline-6- sulphonic acid (ABTS) assays from traditional Chinese herbs is employed as the data set. The aim of the study is to develop reliable quantitative structure-activity relationship also known as QSAR models of flavonoid compounds using the combination of forward stepwise as variable selection method and multiple linear regression (MLR) analysis. The suitable dimensional block of descriptors and significant descriptors that contribute to the antioxidant properties of flavonoids are identified. The performance of the QSAR models are_{2 2 2} reported as r_calc and are validated using cross-validation (r_cv), the external test set (r_pred) and Y-randomisation (r²) to confirm the reliability of the model. Based on the findings, the developed models are robust and reliableand able to explain 78 % variance of antioxidant activity. From the QSAR models, two selected descriptors thatsignificantly affect the antioxidant activity are topological indices (PW5) and 2D-autocorrelations (JGI4). Both of them belong to the 2-dimensional (2D) block of descriptors. This finding proves that the simpler 2D descriptors appear to be sufficient and beneficial information and perform better in building predicted model than 3D descriptors.