Convergence of replica exchange molecular dynamics

Area of Science:

• Computational Chemistry
• Biophysics
• Molecular Dynamics

Background:

• Replica exchange molecular dynamics (REMD) is a generalized-ensemble algorithm designed to overcome local energy traps in simulations.
• Efficient conformational sampling is crucial for understanding molecular behavior, particularly for peptides and proteins.

Purpose of the Study:

• To evaluate the accuracy and efficiency of REMD in reproducing conventional molecular dynamics (MD) results.
• To assess REMD's capability in enhancing conformational sampling for a fast-folding peptide.

Main Methods:

• Two sets of REMD simulations were performed on a 21-amino-acid peptide with differing initial conformations (extended and helical).
• Conventional MD simulations were conducted for comparison, alongside extended MD simulations at nine temperatures.
• Analysis focused on convergence, average helicity, free-energy profiles, and autocorrelation times.

Main Results:

• REMD simulations showed rapid convergence (within 1.0 ns) regardless of initial conformation, unlike conventional MD.
• Excellent agreement between REMD and extended MD results was observed for temperatures above 300 K.
• REMD significantly enhanced sampling efficiency by up to 71.5 times compared to regular MD at various temperatures.

Conclusions:

• REMD accurately and efficiently reproduces long-time MD simulation results, particularly at temperatures >300 K.
• REMD offers substantial improvements in conformational sampling efficiency over conventional MD.
• Further simulation time may be needed for satisfactory convergence at low temperatures (<300 K).