We have studied a small 4-residue protein with CHARMM's
parameter set 22 (with CMAP correction terms) in explicit
TIP3 water (constant pressure & temperature), SHAKE on
hydrogen bonds, 2fs time-step (leap frog).
We note that the average bond lengths are all slightly
above their equilibrium value of the harmonic potential.
For example, the N-terminal bond +H3N---C(alpha) of the
first residue (Gly) has an average value of 1.50 Ang;
whereas the equilibrium bond length of the CHARMM potential
is 1.48 Ang in the parameter set 22.
When we construct from the trajectory the potential of
mean force we obtain a harmonic potential that is
slightly shifted to larger values, and the force
constant is also slightly increased (250 kcal mol-1 Ang-2
instead of 200 kcal mol-1 Ang-2 of the CHARMM potential).
This is not due to statistical errors as you can see by
the error bars; the bars correspond to the standard
deviation for five independent runs (each them 1.5 microseconds long!).
What could be the reason for this increase in
the equilibrium distance and the force constant?
Is this a general behavior?