Using Markov models to simulate electron spin resonance spectra from molecular dynamics trajectories

Area of Science:

• Computational Chemistry
• Biophysics
• Spectroscopy

Background:

• Simulating electron spin resonance (ESR) spectra requires extensive molecular dynamics (MD) data, posing significant computational challenges.
• Generating hundreds of nanosecond-long MD trajectories for spin-labeled proteins is often infeasible.
• Accurate modeling of spin label dynamics is crucial for interpreting ESR spectra.

Purpose of the Study:

• To develop an efficient computational method for simulating ESR spectra from MD data.
• To construct accurate stochastic models of spin label dynamics from atomistic simulations.
• To reduce the computational burden associated with generating long MD trajectories.

Main Methods:

• A two-step procedure based on the hidden Markov model (HMM) framework was employed.
• A discrete-time Markov chain process was built to capture spin label dynamics.
• The HMM was applied to atomistic trajectories of a spin-labeled poly alanine alpha helix in explicit solvent.

Main Results:

• The developed Markov model accurately represents internal spin label dynamics on time scales longer than 150 ps.
• The model provides insights into long-lived conformations of the spin label.
• The methodology successfully generates stochastic trajectories for ESR spectral simulations.

Conclusions:

• Hidden Markov models offer an efficient approach to modeling spin label dynamics for ESR spectroscopy.
• This method significantly reduces the computational cost of simulating ESR spectra from MD.
• The technique is applicable to various spin-labeled systems, including proteins and peptides.