New drug development is one of the most significant scientific challenge related to human welfare. Drugs work in vivo by interacting with bio-molecules in cells especially proteins. The structures of protein and their dynamic functions are the key information to design innovative molecules as drugs. However, until now, the number of U.S. Food and Drug Administration (FDA) approved drugs is 2715 [5]. It is challenging to design new drug-like molecules based on current drugs because of the limited knowledge toward drug structures and drug-protein interactions. Additionally, Alpha-Fold [12], a deep learning algorithm for predicting protein structures, has extended our knowledge of protein structures from 198, 848 to 1,198,209 [1]. To generate more drug molecules based on those protein structures, deep-learning based computational tools are applied.