Identification of ligand templates using local structure alignment for structure-based drug design

Area of Science:

• Computational biology
• Structural bioinformatics
• Drug discovery

Background:

• High-resolution protein-ligand structures are increasing.
• Similar binding sites can share information about their ligands.

Purpose of the Study:

• To develop a tool for identifying structurally similar protein binding sites.
• To use identified binding sites for template-based ligand prediction and design.

Main Methods:

• Developed graph-based local structure alignment (G-LoSA) tool.
• Searched protein-ligand binding-site structure library for geometric and physicochemical similarity.
• Validated G-LoSA against 76 benchmark targets from the Astex diverse set.

Main Results:

• G-LoSA identified a highly similar template ligand from a nonhomologous protein complex in over half of benchmark targets.
• Structural fragments from multiple template ligands can be assembled to design novel ligands.
• Template-based ligand modeling shows potential for de novo ligand design.

Conclusions:

• G-LoSA is effective in identifying structurally similar binding sites for ligand modeling.
• Template-based ligand modeling is a viable strategy for de novo drug design.
• This method complements receptor structure-based approaches in drug discovery.