Middle-way flexible docking: Pose prediction using mixed-resolution Monte Carlo in estrogen receptor α

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Simulating protein-ligand interactions faces a trade-off between speed and accuracy.
• Traditional docking methods sacrifice protein flexibility for speed, leading to inaccuracies.
• Fully flexible molecular dynamics simulations are computationally expensive and have limited sampling.

Purpose of the Study:

• To develop a flexible docking approach (mixed-resolution Monte Carlo - MRMC) balancing speed, protein flexibility, and sampling power.
• To assess the role of receptor flexibility, non-equilibrium candidate Monte Carlo (NCMC), and pose-clustering in docking performance.
• To evaluate the MRMC approach for flexible or poorly resolved protein targets, using estrogen receptor α as a case study.

Main Methods:

• Developed a mixed-resolution Monte Carlo (MRMC) docking approach.
• Treated the protein binding region atomistically and the rest of the protein with a Gō model for flexibility.
• Used implicit solvation and examined cross-docking pose prediction, NCMC, and pose-clustering for scoring.
• Compared MRMC performance against Autodock smina using 61 estrogen receptor α ligands.

Main Results:

• Incorporating protein backbone flexibility significantly improved docking energies and protein-ligand interactions.
• The MRMC approach demonstrated improved performance compared to Autodock smina.
• Non-equilibrium candidate Monte Carlo (NCMC) provided only modest improvements in ligand pose sampling.
• Energy-ranked poses showed improvement with flexibility, but cluster information and NCMC did not yield significant gains.

Conclusions:

• Protein flexibility, particularly backbone flexibility, is crucial for accurate protein-ligand docking.
• The MRMC approach offers a promising balance between computational cost and simulation accuracy for flexible targets.
• Further refinements in force fields, solvation, and NCMC moves could enhance the MRMC model's capabilities.