LigandForge: A Web Server for Structure-Guided De Novo Drug Design

Area of Science:

• Computational chemistry
• Drug discovery
• Bioinformatics

Background:

• De novo drug design is complex, requiring specialized expertise and incurring high costs.
• Existing computational methods often lack accessibility for researchers without programming backgrounds.

Purpose of the Study:

• To present LigandForge, a user-friendly webserver for structure-guided de novo ligand generation.
• To overcome barriers of cost and technical expertise in computational drug discovery.

Main Methods:

• Utilizes a structure-guided framework for assembling molecules from fragment libraries.
• Incorporates voxel-based property grids for spatial mapping of molecular properties.
• Employs reinforcement learning and genetic algorithms for lead optimization and refinement.

Main Results:

• LigandForge integrates structural validation, binding-site characterization, and retrosynthetic feasibility analysis.
• Generated ligands adhere to physicochemical constraints like molecular weight and LogP.
• The platform provides real-time 3D visualization and synthesis-aware environment.

Conclusions:

• LigandForge democratizes de novo drug design by offering a web-based, accessible platform.
• It effectively bridges the gap between structural data and experimentally feasible lead compounds.
• The tool eliminates the need for local software installation, enhancing usability.