Fourth generation detour matrix-based topological descriptors for QSAR/QSPR - Part-2: application in development of models for prediction of biological activity.

Abstract

Augmented path eccentric connectivity topochemical indices (reported in part-1 of the manuscript) along with 42 diverse non-correlating molecular descriptors (shortlisted from a large pool of 2D and 3D MDs) were successfully utilised for the development of models through decision tree, random forest and moving average analysis for the prediction of antitubercular activity of aza and diazabiphenyl analogues of active compound (6S)-2-Nitro-{[4-(trifluoromethoxy)benzyl]oxy}-6,7-dihydro-5H-imidazo[2,1-b][1,3] oxazine (PA-824). The statistical significance of the proposed models was assessed through overall accuracy of prediction, intercorrelation analysis, sensitivity, specificity and Matthew's correlation coefficient (MCC). The accuracy of prediction of the proposed models varied from a minimum of 81% to a maximum of ∼99%. High accuracy of prediction amalgamated with high MCC values clearly indicates robustness of the proposed models. The said models offer a vast potential for providing lead structures for the development of potent antitubercular drugs.