Abstract
In the present study, an in silico approach using decision tree, random forest and moving average analysis has been applied to a data set comprising of 53 analogues of 5-alkyl-2-alkylamino-6-(2,6-difluorophenylalkyl)-3,4-dihydropyrimidin-4(3$H$)-one for development of models for prediction of anti-HIV-1 activity. A total of 46 2D and 3D molecular descriptors of diverse nature, have been used for decision tree and random forest analysis. The value of majority of these descriptors for each analogue in the dataset was computed using E-Dragon software (version 1.0). An in-house computer program was also employed to calculate additional topological descriptors which were not included in E-Dragon software. Random forest correctly classified the analogues into active and inactive with an accuracy of 85%. A decision tree was also employed for determining the importance of molecular descriptors. The decision tree learned the information from the input data with an accuracy of 98% and correctly predicted the cross-validated (10 fold) data with accuracy up to 77%. The best five descriptors identified by decision tree analysis were subsequently used to build suitable models using moving average analysis. The use of models based upon these non-correlating molecular descriptors resulted in the prediction of anti-HIV-1 activity with an overall accuracy of 83–96%. Moreover, active ranges of the proposed models not only revealed high potency but also exhibited improved safety as indicated by relatively high values of selectivity index. The statistical significance of models/ indices was assessed through intercorrelation analysis, sensitivity, specificity and Matthew's correlation coefficient. High predictability offer proposed models a vast potential for providing lead structures for development of potent but safe anti- HIV-1 agents.