Dynamical coring of Markov state models.

Area of Science:

• Computational chemistry
• Biophysics
• Molecular dynamics simulations

Background:

• Accurate definition of metastable conformational states is crucial for Markov state models (MSMs) in biomolecular dynamics.
• Density-based clustering algorithms can identify microstates but face challenges with projection artifacts and insufficient sampling.
• Misclassification in transition regions leads to artificially short lifetimes of metastable states.

Purpose of the Study:

• To introduce and validate dynamical coring as a method to improve the accuracy of MSMs.
• To address the issue of misclassified transitions in low-dimensional molecular dynamics data.
• To enhance the description of complex biomolecular dynamics, especially for unfolded states.

Main Methods:

• Utilized a density-based geometrical clustering algorithm for dimensionality reduction of molecular dynamics trajectories.
• Implemented dynamical coring, requiring a minimum residence time in a state for transition counting.
• Applied the method to molecular dynamics simulations of alanine dipeptide and villin headpiece.

Main Results:

• Dynamical coring significantly improved the Markovianity of metastable states.
• Chapman-Kolmogorov tests confirmed the enhanced Markovian property.
• Implied time scales of the Markov model were increased, indicating more accurate state lifetimes.

Conclusions:

• The combination of density-based clustering and dynamical coring provides a robust approach for defining metastable states in biomolecular dynamics.
• This method effectively mitigates artifacts from dimensionality reduction and sampling limitations.
• The enhanced MSMs offer high structural and temporal resolution for studying complex dynamics of unfolded biomolecules.