From the NMR data you could deduce a 3D structure with an uncertainty of the angles 4-10o. This structure can be used for MD analysis. Such as analysis, however, is not reliable one due to the great uncertainty of the geometry parameters obtained by NMR. This should affect the energetics by MD drastically. Reliable information is obtained using crystallographic parameters as input.
I'm pretty sure the uncertainty of dihedral angles in X-ray structures is also at least 5 degrees.
https://spin.niddk.nih.gov/bax/lit/508/229.pdf
(see section 'accuracy of X-ray phi angles).
Also if you compare for example 2 BPTI X-ray structures in the same crystal form, the RMS differences in phi angles is ~5 degrees. And that's at a resolution of 1.0 to 1.1 Å. It goes up with worse resolution.