Classification of ligand molecules in PDB with graph match-based structural superposition.

Area of Science:

• Computational chemistry
• Structural bioinformatics
• Cheminformatics

Background:

• Accurate classification of small molecule ligands is crucial for drug discovery and understanding biological processes.
• Existing methods for ligand classification may not fully capture structural nuances, leading to suboptimal grouping.

Purpose of the Study:

• To improve the COMPLIG algorithm for small molecule ligand classification by incorporating structural superposition.
• To re-classify ligands in the Protein Data Bank (PDB) using the enhanced algorithm and compare the results with previous methods.

Main Methods:

• The COMPLIG algorithm was modified to include a structural superposition process for atom-atom matching verification.
• 16,660 small molecule ligands from the PDB were classified based on their 3D structures using the enhanced COMPLIG.

Main Results:

• The enhanced COMPLIG algorithm classified 16,660 PDB ligands into 7561 clusters, compared to 3371 clusters using the previous method.
• The new classification system showed an increase in singleton clusters, highlighting differences in chirality, conformation, substructures, and bond length.
• Superposition-based classification effectively grouped functionally related ligands, such as drugs targeting specific biological processes.

Conclusions:

• Structural superposition significantly refines small molecule ligand classification by enforcing stricter atom-atom matching.
• The enhanced COMPLIG algorithm provides a more detailed and functionally relevant categorization of ligands.
• This improved classification aids in identifying ligands with specific biological activities and structural characteristics.