New POCKET

Research on "POCKET" (the term encompassing various related projects) focuses on improving the efficiency and accuracy of computational methods for analyzing and interacting with protein binding pockets. Current efforts center on developing advanced machine learning models, including diffusion models and graph neural networks, to generate novel ligands, predict binding affinities, and enhance the generalizability of docking algorithms. These advancements aim to accelerate drug discovery by enabling more efficient and accurate virtual screening and de novo ligand design, ultimately leading to the development of new therapeutics. Furthermore, research extends to optimizing model training for resource-constrained environments, improving the efficiency of existing algorithms for time series analysis related to pocket interactions.