Development of an Automatic Pipeline for Participation in the CELPP Challenge

Area of Science:

• Computational Chemistry
• Structural Biology
• Drug Discovery

Background:

• Molecular docking is crucial for Structure-Based Drug Design (SBDD) to predict ligand-receptor binding modes.
• Current docking tools face limitations, including missed correct poses and scoring function inaccuracies, impacting drug design reliability.

Purpose of the Study:

• To present a fully automated pipeline for ligand pose prediction.
• To validate the pipeline's performance in real-world scenarios using the Continuous Evaluation of Ligand Pose Prediction (CELPP) Challenge.

Main Methods:

• Developed an automated pipeline for pose prediction.
• Integrated a strategy to mine and utilize existing data for empirical restraints.
• Guided the docking process using defined empirical restraints.

Main Results:

• The pipeline successfully generated predictions for most targets.
• Achieved poses with low Root-Mean-Square Deviation (RMSD) values compared to crystal structures.
• Demonstrated the pipeline's capability in automated protein-ligand complex prediction.

Conclusions:

• The automated pipeline shows promise for improving ligand pose prediction in drug design.
• The study highlights significant challenges in automatic protein-ligand complex prediction.
• Future pipeline versions will address identified limitations for enhanced performance.