Modeling the Antifungal Activity of Benzenic Compounds Using 2D-QSAR and Advanced Statistical Tools

Our study primarily aims to model the antifungal activity of a series of 28 benzenic compounds using the 2D-QSAR method, as well as various statistical tools such as Principal Component Analysis (PCA), Multiple Linear Regression (MLR), and Multiple Nonlinear Regression. This research is structured around three major steps: data selection, molecular descriptor generation, and the construction and validation of predictive models for the studied activity. To develop and validate our QSAR model, we divided the dataset into two groups: 22 molecules for the training set and 6 molecules for the test set, with a random selection of molecules. Both sets were validated separately using internal and external tests, including y-randomization. The model shows high values for R², R²test, and MSE (R² = 0.721, R²test = 0.747, MSE = 0.079), confirming its reliability.