Understanding how proteins and ligands interact is essential for drug discovery, especially for prolyl-tRNA synthetase (ProRS), which is responsible for attaching proline to the corresponding tRNA molecule, a key step in protein biosynthesis in all living organisms. Thus, species-specific inhibitor design for this target holds a key promise in the development of antibiotics with minimal side effects. In the current study, the binding affinities of ligands as well as protein-ligand interactions have been studied for several ProRSs across different host species. Both the physics-based and machine learning models have been utilized, as the latter group of models are computationally inexpensive. The classical physics-based model predicts the affinities by combining the hydrogen bonding, electrostatic, van der Waals, and implicit solvation, while the machine learning model utilizes a deep learning architecture through graph convolutional neural network stitched to artificial neural network. The latter approach enables a faster and more scalable screening of potential drug candidates. Results obtained from the screening method will be compared against a physics-based simulation of molecular interactions and their corresponding binding affinities for the various ProRS enzymes. This research has the potential to enhance drug discovery by improving the speed and scalability of molecular interaction predictions.
