Development of QSAR Models for Predicting Anticancer Activity of Heterocycles

Abstract

In drug discovery, quantitative structure-activity relationship (QSAR) modelling has become a potent computational method, particularly for the early assessment of heterocyclic compounds' potential for anticancer effects. In order to forecast the anticancer activity of specific heterocycles tested on animal models, this study creates reliable QSAR models using molecular descriptors. The study used a dataset of 60 heterocyclic derivatives that have been shown to have anticancer properties in vivo (in mouse models). Multiple Linear Regression (MLR), Partial Least Squares (PLS), and Support Vector Machine (SVM) techniques were used to construct the models. The models' predictive power was validated by cross-validation and external test set validation. By identifying important structural characteristics that affect anticancer potency, the study opens the door for logical drug design and lessens reliance on animal testing by using virtual screening.