Ligand-guided receptor optimization.

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Receptor models from homology or crystallography often have suboptimal binding pockets for docking and virtual screening.
• Inaccurate models or induced fit bias can hinder drug discovery efforts.
• Known ligand information is crucial for improving receptor model performance.

Purpose of the Study:

• To present the ligand-guided receptor optimization (LiBERO) algorithm.
• To enhance the accuracy of receptor binding pockets for virtual screening.
• To improve the performance of docking and screening applications.

Main Methods:

• Generating energetically feasible protein conformers using normal mode analysis and Monte Carlo sampling.
• Iteratively selecting the best protein conformers based on a fitness function.
• Utilizing small-scale virtual screening tests to quantify conformer performance.
• Implementing the automated algorithm on Linux clusters for parallel scaling.

Main Results:

• The LiBERO algorithm effectively exploits ligand information to select optimal protein models.
• The fitness function quantifies a conformer's ability to identify known ligands.
• The algorithm has demonstrated success in binding pocket optimization and virtual ligand screening.

Conclusions:

• The ligand-guided receptor optimization (LiBERO) algorithm improves protein model accuracy for drug discovery.
• LiBERO enhances virtual screening performance by optimizing receptor binding pockets.
• The algorithm offers a scalable and practical solution for computational drug design.