MedChem Game: Gamification of Drug Design

Abstract

The growing importance of computer methods in drug discovery encourages greater accessibility of cheminformatics tools. The methods that enable automated analysis of molecules, e.g., molecular docking or machine learning models, remain out of reach for nonexpert computer users. To address the shortage of easily accessible tools that can be used to practice drug design, we created MedChem Game, an Android application that uses gamification and artificial intelligence to help users learn about medicinal chemistry and design new small-molecule drugs. Our application includes a simplified molecule drawing tool used to propose new drug candidates that can be docked to one of the four target proteins currently implemented in the game. Additionally, we have implemented a 3D ligand–protein viewer so that players can inspect molecular docking results. All the designed compounds are stored in a database with the sequence of operations the user performs. The key component of our system is a continuously trained deep generative model that utilizes user-developed compounds to improve the quality of generated molecules. Learning directly from the expertise and creativity of players, including medicinal chemists, could help advance computer-aided drug design in the future. Moreover, MedChem Game could be a useful supplement to courses on small-molecule drug design.