Abstract
Drug discovery aims at designing novel molecules with specific desiredproperties for clinical trials. Over past decades, drug discovery anddevelopment have been a costly and time consuming process. Driven by bigchemical data and AI, deep generative models show great potential to acceleratethe drug discovery process. Existing works investigate different deepgenerative frameworks for molecular generation, however, less attention hasbeen paid to the visualization tools to quickly demo and evaluate model'sresults. Here, we propose a visualization framework which provides interactivevisualization tools to visualize molecules generated during the encoding anddecoding process of deep graph generative models, and provide real timemolecular optimization functionalities. Our work tries to empower black box AIdriven drug discovery models with some visual interpretabilities.