Dear Nic(holas), Nicholas Keep wrote:
I have a structure with an unsaturated lipid in the active site, which according to the chemists has one cis and one trans double bond. I would expect these to be controlled by the torsion angles in the cif files.
Yes, and a planar restraint (which is more converges faster).
However altering torsions between 0 and 180 makes no difference to the refinement and I get back what I put in terms of the fit of the molecule in coot.
That may very well be the case if you do not activate torsion angle refinement.
Coot is great for fitting such molecules into the density, but you can flip the cis-trans fairly easily.
Yes, I agree. Coot currently ignores const torsions (which is what I am guessing that you have for your trans bonds) in torsion refinement, I think, this is a mistake.
BTW, It's much harder to cis-trans flip if you use hydrogens, I find.
Is it correct that I just have to make sure the bonds are correct before starting refinement as the energy penalty for a cis/trans bond being wrong despite the torsion being set to a constant is simply too small?
You *can* make coot fix up you cis-trans errors (and indeed prevent them from happening in the first place) but currently the interface is non optimal. You need to hand edit the const trans torsion restraints in the cif file so that the torsion id does not contain "const" and change sigma and period to 0.01 and 1 (say).
(And you need to turn on torsion restraints, obviously (under the Refinement/Regularize Control button).)
Paul.
