Machine Learning–Guided QSAR Screening of Fluconazole Analogs and FDA-Approved Drugs against Candida albicans, with Docking, Molecular Dynamics, and ADMET Analysis

Infections caused by Candida albicans represent an increasing public health concern, with the pathogen classified by the WHO as critically important. Due to fluconazole resistance, this study employed a Machine Learning (ML)-guided QSAR workflow to prioritize analogs and drugs with potential antifungal activity. Regression models (Extra Trees, Random Forest, and Gradient Boosting) showed robust and statistically comparable performance (Friedman test, p  = 0.061). Virtual screening identified ZINC000299869730 as the top candidate. Molecular docking against exo-(1,3)-glucanase (PDB ID: 1EQP) predicted a superior in silico binding affinity (–10.27 kcal/mol) compared to fluconazole (–8.87 kcal/mol), suggesting potentially stronger interactions at the catalytic site. Molecular dynamics simulations (200 ns) confirmed complex stability. Although the ADMET profile was comparable to fluconazole, predicted mutagenicity (AMES) and hepatotoxicity signals were identified. This study demonstrates the effectiveness of QSAR/ML models in prioritizing candidates, highlighting ZINC000299869730 as a computationally prioritized lead. The main limitation remains the lack of experimental validation, emphasizing the need for in vitro assays to confirm the predicted efficacy and safety.