Application of artificial intelligence and machine learning in natural products-based drug discovery

Synergistic acceleration of machine learning and molecular docking for prostate-specific antigen ligand design

“…15 In contrast, ligand-based design encompasses methodologies such as Quantitative Structure-Activity Relationship (QSAR) and pharmacophore modeling, aiming to identify ligand structures exhibiting structural or functional similarities to known ligands 16 The integration of both structure-based and ligandbased virtual screening methodologies has demonstrated notable success in drug development endeavors, thereby enhancing the likelihood of successfully identifying target compounds. 17,18 While direct utilization of open source databases like ZINC enables the screening of a vast array of small molecule compounds with binding potential to target proteins, challenges arise when dealing with unknown proteins. In such cases, the establishment of effective pharmacophore models becomes impractical, thereby necessitating substantial efforts for compound property verication.…”