The following paragrphs were taken from:

 Ensign DL, Kasson PM, Pande VS. "Heterogeneity even at the speed limit of folding: large-
scale molecular dynamics study of a fast-folding variant of the villin headpiece" J. Mol. 
Biol. 374 (3): 806-816. (2007) 

The equilibrated native state was used both as the starting point of native state 
simulations and as the starting point of simulations at 373 K to generate thermally 
denatured structures. The latter were nine 10 ns simulations of 1 fs time steps, starting 
with the native structure, at 373 K, with velocities assigned randomly from a Maxwell–
Boltzmann distribution, and constant system volume. All bond lengths were constrained. 
Otherwise, the parameters were as described for the 300 K equilibration described above. 
The final structures from these 373 K simulations were used as the starting point for 
folding studies (at 300 K). These structures were equilibrated at 300 K for 10 ns (using 2 
fs time steps) at constant volume, with the protein coordinates fixed. During these 
simulations, the long-range electrostatic forces were treated with a reaction field 
assuming a continuum dielectric of 78, and the van der Waals was treated with a switch 
from 0.7 nm to 0.8 nm. The neighborlist was shortened to 0.7 nm in order to improve the 
computational performance of the system.

 A great deal of the native structure was lost during the 373 K simulations, although a 
surprising degree of native-like structure remains in the initial structures. For instance, 
unfolded structures 0, 1, 4, and 7 had C RMSD measures for helix 1 consistent with the 
native state simulations. Two unfolded configurations, 5 and 6, contained none of the 
structure we assess in Tables 1 and 2. None of the unfolded structures had a C root 
mean-square displacement (RMSD) or a number of native contacts consistent with the 
folded simulations.