GalaxyCDock: Webserver for Covalent Protein-Ligand Binding Mode Prediction

Area of Science:

• Computational chemistry
• Drug discovery
• Structural biology

Background:

• Covalent ligands offer unique therapeutic potential by targeting proteins resistant to conventional drugs.
• Accurate prediction of covalent ligand binding is essential for drug specificity and safety.
• Existing computational tools for covalent docking lack accessibility and precision.

Purpose of the Study:

• To introduce GalaxyCDock, a novel web server designed for covalent protein-ligand docking.
• To enhance the accuracy and accessibility of computational tools for covalent ligand binding prediction.
• To provide a practical alternative for modeling covalent interactions in drug design.

Main Methods:

• GalaxyCDock integrates efficient pose sampling from GalaxyDock2 with a deep learning scoring function, GalaxyDock-DL.
• The server predicts binding modes for covalent ligands targeting specific proteins.
• Performance was evaluated on standard and newly curated datasets.

Main Results:

• GalaxyCDock demonstrated superior performance compared to established tools like AutoDock4 and DOCK6.
• High accuracy was achieved in both re-docking (up to 80%) and cross-docking (up to 61%) tasks.
• GalaxyCDock offers a viable alternative to advanced models when receptor structures are known.

Conclusions:

• GalaxyCDock significantly advances covalent protein-ligand docking capabilities.
• The developed tool addresses the limitations of current computational methods.
• GalaxyCDock is publicly accessible, facilitating its use in drug discovery research.