An Automated Workflow for Diagnosing Sampling Issues Caused by Slow Torsional Motions in Molecular Simulations

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Physics-based methods, like binding free energy calculations, are crucial for early-stage drug discovery.
• Accurate calculations depend on thorough sampling of protein-ligand conformations, including rotatable bonds.
• Sampling problems, often due to slow torsional rotations, can lead to inaccurate results and are hard to detect.

Purpose of the Study:

• To develop an automated post-simulation analysis method for diagnosing sampling issues in binding free energy calculations.
• To specifically identify problems caused by slow torsional rotations in ligands and protein side chains.
• To improve the reliability of computational drug discovery by ensuring adequate conformational sampling.

Main Methods:

• Developed an automated method for post-simulation analysis of binding free energy calculations.
• Analyzed dihedral angle states for torsions in ligands and protein residues near the binding site.
• Flagged potential sampling issues based on low transition counts between dihedral angle states.

Main Results:

• The automated method successfully detected sampling issues caused by slow torsional rotations.
• These issues were often subtle and could have been missed by traditional convergence monitoring.
• The identified sampling problems have the potential to significantly impact calculated free energy values.

Conclusions:

• The developed method provides an effective way to automatically diagnose sampling problems in binding free energy calculations.
• This approach enhances the reliability of computational drug discovery by ensuring better conformational sampling.
• Early detection of sampling issues leads to more accurate compound prioritization in drug development.