Abstract
Quantitative structure-activity relationship (QSAR)-based classification approach is one of the important chemometric tools in drug discovery process for categorizing the target protein inhibitors into more active and less active classes. In this background, we have presented here a novel approach of two-fold QSAR-based classification modeling for the Plasmodium falciparum carbonic anhydrase (PfCA) inhibitors using 2D-QSAR and linear discriminant analysis (LDA) methods. The logic of applying this concept is to ensure more accurate classification of compounds and to draw some concrete conclusion about structure-activity relations for further work, in absence of 3D-protein structure and lack of sufficient experimental data using the PfCA target. The 2D-QSAR modeling analysis suggested the importance of electrotopological, electronic, extended topochemical atom, and spatial (Jurs) indices for modeling the inhibitory activity against PfCA. The LDA model analysis showed that spatial (Jurs), electrotopological and thermodynamic indices were the discriminating features to differentiate the inhibitors into more active and less active groups. The classification ability of both the models for training and test sets was checked by different qualitative validation parameters such as sensitivity, specificity, accuracy, recall, precision, F-measure and G-means. The classification results revealed that the developed models were significant in classifying the more active inhibitors as compared to the less active inhibitors of both training and test sets. The structural features unveiled from these two models could be utilized for the selection of more active compounds against PfCA in the database screening process.
Keywords: 2D-QSAR, linear discriminant analysis, Plasmodium falciparum carbonic anhydrase, two-fold classification.
Related Journals
Current Computer-Aided Drug Design
21