Visualizing-Deep-Graph-Generative-Models-for-Drug-Discovery

• Presented at 2020 KDD Workshop on Applied Data Science for Healthcare https://dshealthkdd.github.io/dshealth-2020/

Abstract

• First, we propose a visualization framework which provides interactive visualization tools to visualize molecules generated during the encoding-and-decoding process of deep graph generative models. Also, we provide real-time molecular optimization functionalities.
• Second, we propose an end-to-end de novo drug design approach to generate novel molecules with high binding affinity to a specific target protein. We have conducted some initial experiments to leverage the power of MoFlow[11] (a generative model) and the pre-trained drug-target binding affinity prediction models from DeepPurpose[8]. Our work tries to empower black-box AI-driven drug discovery models with some visual interpretive abilities. We believe our initial exploration of generating target-specific novel drug molecules will provide valuable insights for AI-based approaches to timely combat future unforeseen pandemics.

Paper

• Please see my paper at: https://arxiv.org/abs/2007.10333

Demo: