Unsupervised guided docking of covalently bound ligands

Area of Science:

• Computational chemistry
• Drug discovery
• Structural biology

Background:

• Docking programs are essential for predicting ligand-protein interactions.
• Covalently bound ligands present unique challenges for traditional docking methods.
• Accurate prediction of covalent binding modes is crucial for drug development.

Purpose of the Study:

• To implement and validate a novel approach for covalent docking within the MacDOCK program.
• To assess the performance of MacDOCK in predicting covalent ligand-protein binding modes.
• To evaluate MacDOCK's utility in virtual screening for covalent inhibitors.

Main Methods:

• MacDOCK, a docking program based on DOCK 4.0, was adapted for covalent ligand docking.
• The approach was tested using native and non-native protein structures with known covalent ligands.
• Automated detection of reactive functional groups for covalent docking was incorporated.
• Virtual screening was performed on a molecular database of approximately 11,000 compounds.

Main Results:

• MacDOCK successfully generated accurate binding orientations for covalent ligand-protein complexes.
• The program's performance was comparable to existing docking tools.
• Automated functional group detection facilitated the covalent docking process.
• Virtual screening yielded significant enrichment of known covalent thrombin inhibitors.

Conclusions:

• MacDOCK provides a robust method for docking covalently bound ligands.
• The program demonstrates high efficiency in virtual screening for covalent drug discovery.
• MacDOCK is a valuable tool for identifying novel covalent inhibitors.