Oh Wolfram, you top-poster, you..! On 02/12/2016 17:25, wtempel wrote:
I am again struggling with a similar problem and have only (re?-)discovered Paul's May 23 reply today. Thus, I pushed that "<Cis <-> Trans>" button and, yes, I now can still recognize the proline after real-space refinement. But the COOT planar restraints for the cis-peptide appear to be much weaker than for trans-peptides.
The planar peptide restraints are independent of cis/trans configuration.
It almost feels like the "Cis<->Trans" lifts the trans restraints without imposing cis restraints.
Yes. That right. The trans peptide restraints only work with the configuration was previously trans. There is no cis equivalent.
From the point of view of the peptide: If you were trans and were considering cis, then coot will encourage you to be trans. If you are cis and are considering trans, then coot will let you go ahead without additional intervention.
I have experimented with the "add planar peptide restraints" menu, but did not notice any effect.
I can believe that. That's toggling an additional plane restraints, not a torsion restraint.
Am I going about this problem in the "right" COOT way?
It sounds that you are thinking the right thoughts. I'm not sure about the details for the particular problem though... perhaps a cis-model for a cis pre-Pro that won't stay cis?
Paul.
On Mon, May 23, 2016 at 3:28 PM, Paul Emsley <pems...@mrc-lmb.cam.ac.uk <mailto:pems...@mrc-lmb.cam.ac.uk>> wrote: On 17/05/2016 15:52, wtempel wrote: is it only my perception that COOT handles real-space refinement of Xxx-Pro peptide bonds,(specifically cis peptides?) less gracefully than in the past? It handles it differently, well spotted. If you had a decent or good map and a model that was trans (and it should be cis) [1] then Coot will now have additional restraints to keep it trans that will thwart you. Now you need to explicitly use the Cis <-> Trans tool to fix the problem (and then refine). Why? Because (it seemed to me) that unintentional trans->cis conversion was far more problematic than easy intentional trans->cis was useful. [1] If you ever did it, my validation tutorial had such a case and "solved" the problem beautifully. Paul.