Hi Charles,

it should be as simple as

phenix.geometry_minimization model.pdb ligand.cif

However, note that geometry restraints used in this case are very
simplistic, for example, there are no attraction term. This means
helices may unfold unless you do a good thorough job defining secondary
structure restraints (and generally restrain all known hydrogen bonds).
You can do this by making a hydrogen bond restraint file:

phenix.hbond model.pdb

(model.pdb needs to have explicit H added)

and then supply that file to the above minimization command:

phenix.geometry_minimization model.pdb ligand.cif hbond.eff

It may be a good idea to verify hbond.eff to make sure it is as complete
and accurate as possible. This is because H-bond definitions in this
file are made based on input model and if input model geometry isn't
great, the H-bond annotations (and corresponding restraints) may not be
accurate either.

In summary, here are the steps:

1) Add H to the model:

phenix.ready_set model.pdb

2) Make H-bond restraints definitions:

phenix.hbond model_with_H.pdb

This command will create hbonds_pymol.pml file that you can load into
PyMol and see all H bonds as dashed line. This lets you verify all H
bonds that the program found (or missed). Edit hbond.eff if needed.

3) Finally, run geometry regularization:

phenix.geometry_minimization model.pdb ligand.cif hbond.eff

Let me know if you have any questions! Good luck,

Pavel

On 10/19/22 19:04, CPMAS Chen wrote:
> Hi, All,
>
> I have a structure calculated with NMR restraints but it has quite
> some geometry violation, such as torsion angles, clashes etc.
> The structure has a small molecule ligand bound.
>
> Is it possible to add an CIF file for the ligand and use
> phenix.geometry_minimization to idealize/optimize the protein structure?
>
> I have tried the following and phenix reported an error for missing
> CIF file.
>
> phenix.geometry_minimization model.pdb
> pdb_intepretation.apply_cif_restraints.restraints_file_name=ligand.cif
>
> what would be the correct syntax to add ligand cif for this
> geometry_minimization?
>
> Or what would be the proper way to idealize/optimize a protein
> structure with ligands present?
>
> The structure optimization during NMR calculation is not good enough
> in my case.
>
> Thanks!
>
> Charles