[phenixbb] About Restrain Files for ligands

[Paul Emsley]

On 07/01/2021 18:20, Sharan Karade wrote:
>
>
>  I am stuck with some unusual behavior of restrain files generated by 
> Phenix.ReadySet. I crystallized the protein with ligands which have 
> double bonds between the atoms. As I see clear electron density for 
> the ligand, I drew a ligand molecule in Coot (Ligand builder) and 
> fitted in electron density very nicely. Until here, I can see the 
> double bonds in the modelled ligand.


It is highly likely that you have restraints at this point.


> I generated the restrain files using Phenix.ReadySet to further 
> proceed for refinement.


How?


> Unfortunately, after refinement the double bonds were converted to 
> single bonds.


refinement won't change the bond orders.


> I tried to generate new restrain files and refined the model,


How?


> resulting in similar output.


Similar but different?


> But, fortunately in one case, the double bonds were retained after 
> refinement also. I tried to repeat with other ligands, where I was 
> unsuccessful to retain the double bonds.
>

Hmm... Probably the best way of interacting between Coot and ELBOW at 
the moment is, after you have drawn your ligand in Lidia, click on the 
red-lipped smiley icon to capture the SMILES string - and then feed that 
to ELBOW.


On 07/01/2021 20:03, Pavel Afonine wrote:
>
>
> I'm guessing your are over-relying on graphic representations of your 
> molecules which is discussed here to some extent:
>
> http://phenix-online.org/newsletter/CCN_2016_01.pdf#page=10
>
> What really matters is that the ligand fits the map and its geometry 
> matches what's expected.
>
>

Re Figure 3b: Oh yikes, that doesn't look right - the elements have been 
lost. Can you send me that dictionary so that I can try to reproduce the 
problem?


Paul.