Ligand cluster-based protein network and ePlatton, a multi-target ligand finder

Area of Science:

• Pharmacogenomics
• Cheminformatics
• Structural Biology

Background:

• Small molecules are crucial for cellular communication and physiological processes.
• Traditional drug design often targets a single molecule, but polypharmacology suggests multiple targets may be necessary.
• Advancements in cheminformatics and structural biology enable the design of promiscuous drugs and combination therapies.

Purpose of the Study:

• To analyze protein-ligand interactions and identify patterns of polypharmacology.
• To develop a web platform (ePlatton) for exploring drug-target relationships, pathways, and adverse effects.
• To assist in designing next-generation promiscuous drugs and drug combination therapies.

Main Methods:

• Extracted 234,591 protein-ligand interactions from ChEMBL.
• Constructed and analyzed ligand cluster-based and sequence-based protein networks (LCBN, SBN) using 2D structure similarity.
• Integrated pathway, disease, drug adverse reaction, and target-ligand cluster data into the ePlatton web platform.

Main Results:

• Identified 13,769 ligands interacting with 1477 proteins based on 2D structure similarity.
• LCBN and SBN showed high agreement (normalized mutual information at 0.9).
• Found that lighter ligand clusters were more promiscuous, and highly connected nodes were often protein kinases involved in phosphorylation.

Conclusions:

• ePlatton effectively reduces ligand set redundancy and clarifies compound-target-pathway-side effect relationships.
• The platform demonstrated reliability in validation, with fast performance in virtual screening and information retrieval.
• The findings support leveraging polypharmacology for designing effective drug combinations and next-generation therapeutics.