Diverse classification models for anti-hepatitis C virus activity of thiourea derivatives

Unavailability of vaccine and satisfactory drug regimen has made Hepatitis C a worldwide health issue which needs to be urgently addressed. In the present study, diverse classification models have been developed for anti-Hepatitis C virus (HCV) activity of thiourea derivatives. Values of various molecular descriptors (MDs) were calculated for each of the 121 derivatives through Dragon software. Classification models were developed through decision tree (DT), random forest (RF), artificial neural networks (ANN) and moving average analysis (MAA) using training set comprising 61 derivatives. These models were subsequently validated using test set comprising 60 derivatives. The said models were assessed statistically through intercorrelation, sensitivity, specificity, non-error rate, overall accuracy of prediction and Matthew's correlation coefficient. Non-error rate up to >99% and 82.7% was observed in the training set and test set respectively using aforementioned models. Active ranges of proposed MAA based models exhibited high potency amalgamated with safety as indicated by very low value of EC50 and high value of Selectivity Index. Aforementioned models offer vast potential for providing lead molecules for designing of potent but safe anti-HCV thiourea derivatives. •Models were successfully developed for anti-Hepatitis C activity of thiourea derivatives.•Decision tree, random forest, artificial neural networks and moving average analysis were utilized for development of the said models.•Statistical analysis revealed robustness of the proposed models derived through diverse molecular descriptors.•Proposed models can easily provide lead molecules for potent but safe anti-Hepatitis C virus thiourea derivatives.