Computing long time scale biomolecular dynamics using quasi-stationary distribution kinetic Monte Carlo (QSD-KMC)

Area of Science:

• Computational Biology
• Biophysics
• Chemical Physics

Background:

• Modeling complex biological system dynamics from molecular dynamics (MD) simulations presents significant challenges.
• Markov state models (MSMs) are popular for analyzing MD data but introduce bias due to the Markovian assumption, particularly for complex energy landscapes.
• Non-Markovian dynamics in biomolecular systems limit the accuracy of traditional modeling approaches.

Purpose of the Study:

• To develop a novel computational method that accurately models state-to-state dynamics in complex biological systems, even under non-Markovian conditions.
• To provide a method that retains full time resolution and overcomes systematic bias inherent in Markovian assumptions.
• To introduce a robust approach for analyzing long-time evolution in biomolecular simulations.

Main Methods:

• Developed quasistationary distribution kinetic Monte Carlo (QSD-KMC), a kinetic Monte Carlo approach.
• Incorporated the concept of quasistationary distribution, where dynamics become Markovian after a sufficient dephasing time within a state.
• Utilized dynamical corrections theory to account for correlated events during state transitions.

Main Results:

• QSD-KMC accurately models long-time state-to-state evolution for non-Markovian dynamics.
• The method retains full time resolution, unlike traditional MSMs.
• QSD-KMC generated trajectories are statistically indistinguishable from MD trajectories, even with arbitrary state decomposition.
• The approach allows for Monte Carlo optimization of state boundaries.

Conclusions:

• QSD-KMC offers a significant advancement for modeling biomolecular dynamics, providing accurate long-time predictions.
• The method effectively addresses the limitations of the Markovian assumption in complex systems.
• QSD-KMC is validated on model systems and biomolecular systems, demonstrating its broad applicability.