Antibiotic-resistant Staphylococcus aureus presents a significant global health challenge, necessitating alternative antimicrobial treatments. Essential oils from ethnomedicinal plants offer potential antibacterial activity due to their bioactive compounds. Moreover, possible synergism between these oils leads to improved therapeutic properties. Nevertheless, incorporating elements of traditional medicine into modern pharmaceutical research mandates a molecular-level understanding of such phenomena, as well as means of automation and high-throughput integration. To assess the possibility of such endeavors, our study aims to explore the synergistic and antagonistic effects of these oils using machine learning interfaced with in-silico and in-vitro methods. We apply graph embedding to model interactions between 147 essential oil compounds and use multinomial logistic regression to classify antibacterial effects into synergistic, antagonistic, and neutral categories. In-vitro studies are conducted to confirm the predictions using the minimum inhibitory concentration (MIC) values, and the fractional inhibitory concentration (FIC) indices are used to quantify interactions. We use deep learning-based molecular docking, classical molecular dynamics (MD) simulations, and electronic structure calculations to probe interactions between essential oil compounds and three key proteins vital for S. aureus pathogenicity: Sortase-A, DNA Gyrase, and Dihydrofolate Reductase. This study provides insight into integrating machine learning with in-silico and in-vitro approaches to study synergism, offering a comprehensive framework for evaluating the antibacterial potential of essential oils, supporting their role as promising alternatives to conventional antibiotics against S. aureus.

Anton Perera

University of Kentucky College of Arts and Sciences

Appearances