Validation of DBFOLD: An efficient algorithm for computing folding pathways of complex proteins.

Area of Science:

• Computational Biology
• Biophysics
• Protein Dynamics

Background:

• Atomistic simulations offer insights into protein folding but are computationally limited for large proteins.
• Existing native-centric models often neglect crucial non-native interactions.
• Understanding protein folding is vital for comprehending biological function and disease.

Purpose of the Study:

• To develop a novel algorithm (DBFOLD) for predicting folding pathways of diverse proteins, including non-native interactions.
• To enable prediction of transition rates within protein folding pathways.
• To overcome computational limitations of existing ab initio methods for larger proteins.

Main Methods:

• Combined equilibrium Monte Carlo simulations with enhanced sampling and high-temperature unfolding simulations.
• Jointly analyzed simulation trajectories to compute folding rates using detailed balance.
• Inferred free energies and extrapolated transition rates to physiological temperatures.

Main Results:

• Accurately predicted folding pathways and rates for Streptococcal protein G.
• Significantly reduced computation time for large, misfolding-prone proteins.
• Demonstrated the ability to model the role of non-native intermediates in folding.

Conclusions:

• DBFOLD provides a computationally efficient method for atomistic protein folding simulations.
• The algorithm accurately models folding pathways and rates, including non-native contacts.
• DBFOLD advances the study of protein folding mechanisms relevant to health and disease.