Fragment Pose Prediction Using Non-equilibrium Candidate Monte Carlo and Molecular Dynamics Simulations

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Understanding molecular binding to target proteins is crucial for early-stage drug discovery.
• Accurate prediction of binding modes aids chemists in designing improved molecules with higher binding affinities.

Purpose of the Study:

• To compare the performance of docking, molecular dynamics (MD), and non-equilibrium candidate Monte Carlo (NCMC) + MD methods in predicting binding modes.
• To evaluate the ability of these methods to identify dominant and additional binding modes for fragment-like molecules.
• To compare predicted binding modes with experimental X-ray crystal structures.

Main Methods:

• Docking of 12 fragment-like molecules into the apo-protein crystal structure of soluble epoxide hydrolase.
• Running molecular dynamics (MD) simulations up to 1 microsecond for each molecule.
• Employing a hybrid non-equilibrium candidate Monte Carlo (NCMC) + MD simulation approach.
• Building Markov State Models to define stable ligand binding modes.

Main Results:

• Microsecond MD simulations failed to sample all observed crystallographic binding modes for small, fragment-like molecules.
• The NCMC+MD simulation approach demonstrated improved success in sampling crystal structures.
• The NCMC+MD method achieved better population sampling of binding modes compared to traditional MD.

Conclusions:

• Traditional microsecond MD simulations may be insufficient for fully sampling binding modes of small molecules.
• The NCMC+MD hybrid method offers a more effective strategy for accurately predicting ligand binding modes and populations.
• This study highlights the utility of advanced computational methods in refining drug discovery processes.