Dear Philippe,

I'm not sure it will help and frankly, I don't have any better idea, but here is how one can change the weight of secondary structure restraints.

Before starting, I would check the .geo file to make sure the model deviates quite significantly from the target in the restraints you are going to adjust. This will also give an idea what to adjust.

First parameter to play with is secondary_structure.nucleic_acid.scale_bonds_sigma. Default is 1. All sigmas for h-bond length will be multiplied by this number. The smaller number is tighter restraints.

Next, I would switch parallelity restraints for basepares to planarity. This should make them really planar. This can be achieved by nucleic_acid.base_pair.restrain_planarity=True, nucleic_acid.base_pair.restrain_parallelity=False parameters for individual basepairs. They can be changed in the GUI for individual base pairs in the selection editor, see screenshot below.

If you prefer command-line, or working with parameter file, you can run
phenix.secondary_structure_restraints model.pdb file.param

with file.param containing for example (changing parallelity to planarity):

secondary_structure {
  enabled = True
  nucleic_acid {
    base_pair {
      restrain_planarity = True
      restrain_parallelity = True
      planarity_sigma = 0.2
  }
}

This will produce a parameter file describing SS restraints which can be edited further. For example, you can define planarity_sigma. 

For the sake of completeness, the parameters are listed here: https://phenix-online.org/documentation/reference/secondary_structure.html

Hope it will help,

Best regards,
Oleg Sobolev.

}Screenshot 2024-01-31 at 09.41.42.png

On Tue, Jan 30, 2024 at 10:49 PM CUNIASSE Philippe <Philippe.CUNIASSE@cea.fr> wrote:

Dear Oleg,

sorry for the log.

The strange thing is that the input structure (output from NAMD MDFF where we also apply SS restraints to NA has a correct and better SS than the one at the end of the RealSpace refine.

I also checked the restraints in the .geo file and it seems correct. I only found in previious calculations that the atom identity of some atoms (in the charmm 36 topology) must be changed for the NA restraints to be correctly set by phenix (C3M of THY must be changed for C7). 

My only hypothesis at the moment is that the occupancy refinement (currently at a value of 1.00 for all atoms) tends to distort strongly the extremities of the duplex because the electron density is weak in these regions with a bad shape and that this (presumlably) distorts the NA structure to fit the poor density that does not ressemble the one for a B-DNA in these regions. This could be solved if we have a way to increase the harmonic constants for the SS restraints, but i dont see this possibility.

Your advice on this hypothesis is welcome.

Best regards.

Philippe.


Philippe Cuniasse, PhD/HDR.
Institute for Integrative Biology of the Cell (I2BC)
UMR 9198 CNRS-CEA-Univ Paris Sud
Bat 144 CE-Saclay
91191 Gif-sur-Yvette Cedex, France

Tel: (33) 1 69 08 56 35
Fax: (33) 1 69 08 47 12
Email: philippe.cuniasse@cea.fr <mailto:philippe.cuniasse@cea.fr>
Web: http://biodev.cea.fr/rasmot3d/ <http://biodev.cea.fr/rasmot3d/>
Web: https://www.i2bc.paris-saclay.fr <https://www.i2bc.paris-saclay.fr>

-----------------------------------------------


De : Oleg Sobolev <osobolev@lbl.gov>
Envoyé : mardi 30 janvier 2024 18:21:02
À : CUNIASSE Philippe
Cc : phenixbb@phenix-online.org
Objet : Re: [phenixbb] RealSpaceRefine Nucleic acid SS restraints ?
 
Dear Philippe,

Most often the problem is that the geometry of the input model is so distorted that the procedure fails to find base pairs and therefore restrain them. 

Couple of ways to check for this are: 
1. Look in the .log into 
  Finding SS restraints...
    Secondary structure from input PDB file:
      0 helices and 0 sheets defined
      0.0% alpha, 0.0% beta
      12 base pairs and 21 stacking pairs defined.
    Time for finding SS restraints: 0.04
  Creating SS restraints...

    No hydrogen bonds defined for protein.
    Restraints generated for nucleic acids:
      32 hydrogen bonds
      64 hydrogen bond angles
      0 basepair planarities
      12 basepair parallelities
      21 stacking parallelities

This part would give a general idea if anything worked at all.

2. Inspect .geo file after refinement and look for "Basepair parallelity restraints" and "Bond-like restraints" (secondary structure H-bonds) sections. Here you can see exactly what restraints were applied and if the basepairs in question were in fact restrained.

If you supply the restraints in the .parameter file, check the nucleic_acid.hbond_distance_cutoff parameter. The default is 3.4, you may need to increase it so your annotations are not filtered out.

I'm happy to look at your files (model and parameter, if any) to figure out what is happening. Please send them off-list indicating problematic residues. Files will be treated confidentially.

Best regards,
Oleg Sobolev.

On Tue, Jan 30, 2024 at 7:20 AM CUNIASSE Philippe <Philippe.CUNIASSE@cea.fr> wrote:

Dear all,

I am currently refining a protein-nucleic acid structure obtained by MDFF (NAMD) with a 2.9 Å cryo-EM map.

Dues to the poor density of the map at the end of the DNA duplex (likely due to breathing at the ends of the duplex), the structure of the DNA tends to be distorted.

I tried to set up a set of nucleic acid restraints (Base pairing and base stacking for the full sequence), however, despite the fact that the syntax was checked and no error message is present in the log file, it seems that the restraints are not taken into account as if the constant restraints were two weak. And indeed when examining the refined structure, the pairing and stacking restraints for the base pairs at the extremities of the DNA duplex are not fulfilled.

Have you a suggestion to solve this problem ?

Thanks in advance for your help.

Best regards.

Philippe.


-----------------------------------------------

Philippe Cuniasse, PhD/HDR.

Institute for Integrative Biology of the Cell (I2BC)

UMR 9198 CNRS-CEA-Univ Paris Sud

Bat 144 CE-Saclay

91191 Gif-sur-Yvette Cedex, France

 

Tel:      (33) 1 69 08 56 35

Fax:      (33) 1 69 08 47 12

Email: philippe.cuniasse@cea.fr

Web: http://biodev.cea.fr/rasmot3d/

Web: https://www.i2bc.paris-saclay.fr

signature_2602492259


_______________________________________________
phenixbb mailing list
phenixbb@phenix-online.org
http://phenix-online.org/mailman/listinfo/phenixbb
Unsubscribe: phenixbb-leave@phenix-online.org