Overcoming entropic barrier with coupled sampling at dual resolutions

Area of Science:

• Computational Biology
• Biophysics
• Molecular Dynamics

Background:

• Protein folding is a complex process crucial for biological function.
• Simulating ab initio protein folding is computationally challenging due to vast conformational space.
• Existing methods like replica-exchange struggle with entropic barriers in high-resolution models.

Purpose of the Study:

• To develop an enhanced sampling method for efficient ab initio protein folding simulations.
• To overcome the entropic barrier encountered with high-resolution models.
• To generate accurate Boltzmann distributions in both high and low-resolution simulations.

Main Methods:

• Coupling a high-resolution model (e.g., all-atom force field) with a low-resolution model.
• Ensuring detailed balance condition for Boltzmann distribution generation.
• Testing on analytical energy functions and beta-hairpin peptide folding simulations.

Main Results:

• The new method demonstrates higher efficiency compared to the replica-exchange method.
• Fewer energy calculations are needed to find global minima and converge energies.
• The method explores conformational space more rapidly, confirmed by ergodic measures.
• Imperfect low-resolution models show decreased, but still reasonable, efficiency gains.

Conclusions:

• The enhanced sampling method significantly accelerates ab initio protein folding simulations.
• It effectively navigates the complex protein conformational landscape.
• The method achieves accurate structural distributions comparable to equilibrium simulations.