Accurate ligand-protein docking in CASP15 using the ClusPro LigTBM server

Area of Science:

• Computational biology
• Structural bioinformatics
• Drug discovery

Background:

• The CASP15 competition focused on predicting small molecule-protein interactions.
• Protein models were provided as amino acid sequences or generated by AlphaFold2.
• Accurate ligand binding prediction is crucial for drug development.

Purpose of the Study:

• To evaluate the performance of template-based ligand docking methods in CASP15.
• To compare template-based docking with direct docking on AlphaFold2 models.
• To identify effective strategies for ligand binding site prediction.

Main Methods:

• Utilized ClusPro ligTBM, a template-based ligand docking program, for most targets.
• Employed Glide for direct docking when templates were unavailable.
• Required manual interventions and multiple templates for complex targets.

Main Results:

• The ClusPro ligTBM server was a highly useful tool in the CASP15 ligand prediction category.
• The research group ranked among the top five performing teams.
• All top-performing groups successfully employed template-based docking methods.

Conclusions:

• Template-based docking is a robust method for predicting ligand binding sites, even with AlphaFold2 models.
• AlphaFold2 models, while accurate in backbone prediction, exhibit local differences challenging direct docking.
• Homology-based docking effectively overcomes limitations associated with direct docking on predicted protein structures.