Integrative Machine Learning and Network Pharmacology Approach to Uncover Phytochemical Therapeutics for Melanoma and Breast Cancer

INTRODUCTION: Melanoma and Breast cancer are the leading cancers that cause thousands of deaths annually worldwide. Although several early identification methods and treatments are being used, they have limitations, such as drug resistance and off-target effects. METHODS: To find dual-acting inhibitors for the MDA-MB-435 lineage, we describe a novel integrative approach that combines bioinformatics-driven phytochemical screening with machine learning-based QSAR modeling. This will enable prioritizing natural chemicals with therapeutic promise against both diseases quickly and economically. RESULTS: Among all the models developed, the Extra Trees and Random Forest classifiers performed well, achieving 95% accuracy in identifying the bioactivity class (IC50) of phytochemicals. Furthermore, molecular docking studies revealed that Ombuin outperformed all other test ligands, with docking scores of -5.5 kcal/mol for IL6 and -6.8 kcal/mol for JUN. MDS were performed for the IL6_Ombuin, JUN_Ombuin, IL6_Reference, and JUN_Reference complexes to better understand these interactions. Analysis of RMSD, RMSF, RG, SASA, and Hydrogen bonds for IL6_Ombuin showed lower RMSD and RMSF than for the reference ligand, indicating greater stability. DISCUSSION: The IL6_Ombuin complex exhibited consistent hydrogen bond formation, indicating better structural stability. CONCLUSION: The study's complete pipeline identified Ombuin as a potential hit compound. In the future, in vitro and in vivo studies should be conducted to confirm its therapeutic potential against melanoma and breast cancer.