Docking ligands onto binding site representations derived from proteins built by homology modelling

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Genome projects yield abundant sequence data, identifying novel protein targets for drug design.
• Homology models lack the accuracy for standard drug design tools.
• Accurate protein structural models are crucial for effective drug discovery.

Purpose of the Study:

• To develop a novel computational approach, DragHome, for docking ligands into approximate protein homology models.
• To integrate homology modeling with 3D quantitative structure-activity relationship (QSAR) data for enhanced drug design.
• To validate the DragHome method using known protein-ligand interactions.

Main Methods:

• DragHome analyzes protein binding sites and ligands, translating them into functional descriptions using Gaussian functions.
• Docking is achieved by optimizing the overlap between these functional descriptions.
• Ligand molecular flexibility is incorporated using pre-calculated conformers.

Main Results:

• DragHome successfully docked distinct thrombin inhibitors onto homology models with high accuracy (1.4 Å RMSD).
• The method ranked near-native ligand binding modes as the best solutions.
• DragHome facilitated the automatic alignment of 88 thrombin inhibitors for 3D QSAR model derivation.

Conclusions:

• DragHome provides a robust method for ligand docking into approximate protein models, advancing drug design for novel targets.
• The approach aids in identifying model deficiencies, potentially guiding future protein model refinement.
• DragHome enhances the development of 3D QSAR models, crucial for structure-based drug discovery.