Nathaniel Echols wrote:

On Mon, Jul 15, 2013 at 6:00 PM, Edward A. Berry wrote:

...

REMARK 3 DEVIATIONS FROM IDEAL VALUES. REMARK 3 RMSD MAX COUNT REMARK 3 DIHEDRAL : 12.902 168.987 3277 ^^^^^^^ Could this be my cis peptide, or do I have a really bad dihedral somewhere else?

I would assume the latter, since the deviation of the cis-peptide is only 1.58 degrees, but I think the .geo file should make this clear - the dihedral angles will be sorted by residual (in descending order), and these are coming from the same source as the statistics in the PDB header.

Thanks, got it- its a ligand. The cif file fdoesn't allow a 2-fold rotation that it should. "sinusoidal 1" should be "sinusoidal 2", I guess. or change 90 to -90 for this structure? Interestingly the incorrect restraint doesn't keep it from fitting the density beautifully.

Pavel Afonine wrote:

Oh, makes it hard to resist to re-post this old link to slides #17-18 here: http://www.phenix-online.org/presentations/latest/pavel_validation.pdf

The reason you could get away with this ok is because weights data/restraints used in phenix.refine are global.

Pavel

OK, makes sense. And the sigma was high and weight low for that angle, allowing the clear xray data to overrule the restraint and give the right structure? dihedral pdb=" C10 FAD A1001 " pdb=" N10 FAD A1001 " pdb=" C1' FAD A1001 " pdb=" C2' FAD A1001 " ideal model delta sinusoidal sigma weight residual 90.09 -79.50 169.58 1 2.00e+01 2.50e-03 4.76e+01 (The FAD.cif is from ccp4)