The Landscape of Anti-HBV Activity in Essential Oils: A Machine Learning-Based Virtual Screening Framework and...

Innovative and curative therapies for HBV infection are urgently needed. Essential oils (EOs) offer potential as natural therapeutic alternatives for modulating inflammation and antiviral activity. Extensive screening of anti-HBV EOs promises to yield drug candidates that suppress HBV immune evasion via the modulation of inflammatory pathways, thereby providing new attempts at HBV clearance. This study aimed to investigate the landscape of anti-HBV EOs and develop a virtual screening model to predict the efficacy of EOs against HBV. 292 plant samples were collected and analyzed in terms of extracting their EOs, determining their compositions via GC-MS, and testing their anti-HBV activity. Of the 292 EOs, five EOs exhibited significant anti-HBV activity, including those from Acorus tatarinowii Schott (ATEOs), Caryopteris forrestii, and Vitex agnus-castus. To develop a machine learning model for predicting anti-HBV EOs, we randomly selected 274 EOs to develop a virtual screening and prediction model using Random Forest (RF) and Support Vector Machine (SVM) algorithms. The remaining 18 EOs were used to validate the precision and accuracy of the model. Our machine learning models achieved high precision in predicting anti-HBsAg activity and cytotoxicity. ATEOs emerged as the most potent candidates, with α- and β-asarones identified as the primary active components responsible for the anti-HBV activities. ATEO exerted multiple functions in regulating the MAPK signaling pathway, suggesting multiple mechanisms in modulating inflammatory responses to alter HBV replication. These findings suggested that EOs promise to offer insights into potential aromatherapy applications against HBV. ATEOs exerted anti-HBV activity primarily by modulating the MAPK signaling pathway, making them potent candidates for anti-HBV aromatherapy targeting inflammatory pathway regulation to block persistent HBV infection.