Synthesis, characterization and machine learning prediction of novel quinolone derivatives

The potential applications of quinolones in drug design are growing rapidly. The evolution of synthetic methodologies has enabled the development of new quinolone-based motifs. In the present study, we attempted to introduce various pharmacophoric groups in order to enhance the pharmacological potential of quinolones. In the present work, novel quinolone derivatives containing 1,3,4-oxadiazole-2-thiol, acylhydrazone and pyrazolone moieties (4, 5 and 6) were synthesized and characterized by NMR spectroscopy, FT-IR, and high-resolution mass spectrometry. A dataset of quinolone derivatives with known antimicrobial activity against five microbial strains was used to develop predictive models using four machine learning algorithms: AdaBoost Regressor, Support Vector Regression (SVR), Decision Tree and Random Forest. The best performing model for each strain was selected and used to evaluate newly synthesized compounds. Compounds with strong predicted activity were further analyzed through molecular docking to assess their binding with microbial targets. ADMET properties were also predicted to evaluate drug-likeness and safety, supporting the identification of promising antimicrobial candidates for further testing.