Monte Carlo vs molecular dynamics for all-atom polypeptide folding simulations

Area of Science:

• Computational Biology
• Biophysics
• Statistical Mechanics

Background:

• Accurate protein folding simulations are crucial for understanding biological function.
• Monte Carlo (MC) and Molecular Dynamics (MD) are key computational methods for simulating protein folding.
• The "concerted rotations with flexible bond angles" (CRA) algorithm enhances MC efficiency.

Purpose of the Study:

• To directly compare the efficiency and accuracy of MC and MD algorithms in all-atom polypeptide folding simulations.
• To validate the "concerted rotations with flexible bond angles" (CRA) MC algorithm's performance.
• To assess the reliability of thermodynamic and dynamic properties obtained from both simulation methods.

Main Methods:

• All-atom statistical mechanics folding simulations of three small polypeptides (trpzip2/H1/Trp-cage).
• Direct comparison of a Monte Carlo (MC) algorithm with concerted rotations against Molecular Dynamics (MD).
• Extensive sampling achieved: approximately 10^11 MC configurations and 8 microseconds of MD simulation time.

Main Results:

• Both MC and MD simulations successfully identified the experimentally determined native states for all three polypeptides.
• MC simulations demonstrated 2-2.5 times faster folding compared to MD simulations.
• Thermodynamic and dynamic properties were reliably obtained using both MC and MD, indicating algorithm independence.

Conclusions:

• The efficient MC algorithm, particularly with concerted rotations, provides a faster route to predicting native protein structures.
• MC and MD yield comparable results for thermodynamic and dynamic properties in small polypeptide folding.
• The simplicity and efficiency of MC make it suitable for larger systems and integration with advanced algorithms like replica exchange.