Machine Learning models for prioritizing the synthesis of xanthone derivatives as promising antibacterial agents against methicillin-resistant Staphylococcus aureus

Kielmeyera coriacea is a medicinal plant known in Brazil as “pau-santo”, which present antimicrobial, antitumor, antimalarial and anti-inflammatory properties1. These biological activities could be attributed to the abundance of xanthonic derivatives in this genus2. In this study, we developed a machine learning model based on xanthones with activity against Staphylococcus aureus prioritize compounds for synthesis and biological testing. The dataset of compounds was gathered from ChEMBL database and curated and standardized according to good practices for QSAR modeling3. Then, the machine learning model was used to screen all xanthones commercially available in e-molecules database and two in-house designed databases of xanthones. We predicted their activities and seven xanthones were selected as the most promising virtual hits. Two of them were extracted and isolated from the Garcinia mangostana4 and also will be used as precursor of the other five semi-synthetic xanthones, which will be evaluated against methicillin resistant S. aureus.