Uncertainty quantification for Markov state models of biomolecules constructed using rare event acceleration...

Area of Science:

• Computational Biology
• Biomolecular Simulations
• Statistical Mechanics

Background:

• Markov state models (MSMs) are crucial for analyzing biomolecular dynamics from molecular dynamics (MD) simulations.
• Incomplete sampling in MD trajectories can lead to uncertainties in constructed MSMs.
• Quantifying and reducing uncertainty in MSMs is essential for reliable biological insights.

Purpose of the Study:

• To investigate the effectiveness of rare-event acceleration techniques in reducing MSM uncertainty.
• To demonstrate the application of steered molecular dynamics (sMD) for MSM construction.
• To develop methods for quantifying uncertainty bounds in MSMs derived from accelerated simulations.

Main Methods:

• Utilized deca-alanine as a model system for testing computational methods.
• Employed rare-event acceleration techniques, specifically steered MD, to enhance sampling of biomolecular pathways.
• Developed a framework to derive upper and lower bounds for MSM uncertainty.
• Incorporated safeguards to address potential limitations of acceleration techniques in discovering all relevant pathways.

Main Results:

• Rare-event acceleration techniques significantly reduce MSM uncertainty with high computational efficiency.
• Steered MD successfully contributed to the construction of more complete MSMs.
• The developed approach provides quantifiable uncertainty estimates for the resulting MSMs.
• Safeguards were effective in handling cases where acceleration missed critical pathways.

Conclusions:

• Rare-event acceleration techniques are a computationally efficient strategy for improving the accuracy of MSMs.
• Steered MD offers a viable method for enhancing biomolecular pathway sampling in MSM construction.
• The proposed uncertainty quantification framework enhances the reliability of MSMs derived from accelerated simulations.