Molecule generation toward target protein (SARS-CoV-2) using reinforcement learning-based graph neural network via...

Area of Science:

• Computational Chemistry
• Artificial Intelligence in Drug Discovery
• Molecular Modeling

Background:

• AI accelerates drug discovery by predicting molecule-protein binding affinity.
• Generating novel drug candidates with desired properties (e.g., drug-likeness, target activity) while respecting chemical laws is a significant challenge.
• Developing effective therapeutics against SARS-CoV-2 remains a critical area of research.

Purpose of the Study:

• To develop a graph-based deep learning framework for generating novel therapeutic drug candidates targeting the SARS-CoV-2 protein.
• To enhance molecular properties such as Quantitative Estimate of Drug-likeness (QED) and Dopamine Receptor D2 activity (DRD2).
• To predict binding affinity scores for generated compounds against SARS-CoV-2 targets.

Main Methods:

• A novel reinforcement learning (RL)-based graph generative module utilizing a gated graph neural network (GGNN) and a custom knowledge graph (KG).
• A graph early fusion approach (GEFA) for predicting binding affinity between generated molecules and target proteins.
• Fine-tuning the GGNN model within an RL environment to optimize molecular properties and using KG for efficient molecule screening.

Main Results:

• The framework generated a significant number of valid and unique compounds with enhanced desired properties.
• KG-based screening effectively reduced the search space of candidate molecules while retaining promising binders.
• Achieved a high binding affinity score of 8.185 for a top-ranked generated compound against SARS-CoV-2 3C-like protease (3CLpro).

Conclusions:

• The proposed AI framework successfully generates novel drug-like molecules with high binding affinity for SARS-CoV-2 targets.
• The integration of RL, GGNN, KG, and GEFA offers a powerful approach for accelerating drug discovery and development.
• The generated compounds show potential for further investigation as therapeutic agents against SARS-CoV-2.