Constructing Generalized Sample Transition Probabilities with Biased Simulations.

Area of Science:

• Computational Chemistry
• Biophysics
• Statistical Mechanics

Background:

• Molecular dynamics (MD) simulations are vital for studying molecular kinetics, including reaction paths and rates.
• Enhanced sampling techniques accelerate MD simulations but introduce biases, complicating kinetic analysis.
• Existing methods like diffusion maps struggle with biased data due to altered probability distributions.

Purpose of the Study:

• To develop a method for estimating intrinsic transition probabilities from biased molecular dynamics simulations.
• To overcome the limitations of current techniques in kinetic analysis of complex systems.
• To provide a general framework for recovering unbiased kinetic information.

Main Methods:

• A coarse-grained Markov chain model is employed to estimate pairwise transition probabilities.
• The generalized sample transition probability (GSTP) method is proposed.
• GSTP does not require an underlying stochastic process or kernel function specification.

Main Results:

• GSTP successfully recovers unbiased eigenvalues and eigenstates from biased simulation data.
• Validation was performed on diverse model systems: harmonic oscillator, Müller-Brown potential, alanine dipeptide, and met-enkephalin.
• The method demonstrates robustness across different molecular systems and environments.

Conclusions:

• GSTP offers a robust approach for kinetic analysis of complex systems using biased simulations.
• This method enables accurate determination of transition probabilities, crucial for understanding reaction dynamics.
• GSTP provides a valuable tool for extending the accessible timescales in molecular simulations.