Understanding how a protein folds is a long-standing challenge in modern science.
We have used an optimized atomistic model (united-residue force field) to simulate folding
of small proteins of various structures: HP-36 (alpha protein),
protein A (alpha), 1fsd (alpha+beta) and betanova (beta).
Extensive Monte Carlo folding simulations
(ten independent runs with 10^9 Monte Carlo steps at a temperature)
starting from non-native conformations are carried out for each protein.
In all cases, proteins fold into their native-like conformations
at appropriate temperatures, and
glassy transitions occur at low temperatures.
To investigate early folding trajectories, 200 independent runs with
10^6 Monte Carlo steps are also performed
at a fixed temperature for a protein.
There are a variety of possible pathways
during non-equilibrium early processes (fast process, about 10^4 Monte Carlo steps).
Finally, these pathways converge to the point unique for each protein.
The convergence point of the early folding pathways
can be determined only by direct folding simulations.
The free energy surface, an equilibrium thermodynamic property, dictates the
rest of the folding (slow process, about 10^8 Monte Carlo steps)