Re: [phenixbb] Incorrect secondary structure assignment in a .pdb

Thank you for clarification Ricardo and Pavel. It still amazes me though that different pdb viewers would make such ridiculous SS predictions/assignments at times for a file that does not come with that information in the header. And it seems that during a deposition to PDB I have to manually add the location of helices/sheets into the .cif files generated by the deposition system...otherwise I go for deposition somewhat blind (this has of course nothing to do with Phenix, but adds more to my confusion). Maciej On Thu, Jun 28, 2018 at 4:16 PM, Pavel Afonine wrote:

The point is that no automated secondary structure (SS) annotation software can assign SS fully correctly simply because it relies on input model geometry, and models that are subject to refinement are rarely perfect for these software to do a great job. This is why the Manual (see previous email) recommends getting initial automated assignment first using phenix.secondary_structure_restraints (which uses various annotation methods, including ksdssp) and then manually checking and editing it if necessary.

Tools within Phenix always preserve HELIX/SHEET records in PDB files. Let us know if you have an example where this is not the case.

Pavel

On 6/28/18 13:10, Ricardo Righetto wrote:

Hi Maciej,

If the PDB file has no secondary structure assignments defined in the header and you open it in Chimera, it automatically runs ksdssp in the background to assign helices and sheets, which may or may not be correct and might make you think that you have secondary structure records in the header when you actually have nothing: https://plato.cgl.ucsf.edu/chimera/docs/UsersGuide/midas/ksdssp.html

If in turn you save the PDB from Chimera with such automatically generated assignments, the records corresponding to this assignment will be written to the header. And if it's written in the header, by default phenix.real_space_refine will use these restraints.

Bottom line is that throughout modelling and refinement you should always inspect your output PDB file from whatever program you use (PHENIX, Coot, Chimera, etc) to make sure that the information you think should be in the header is indeed there. I agree that some programs might unnecessarily "get rid" of the secondary structure records in the header and this is a bit annoying.

Hope this helps,

-- Ricardo Diogo Righetto

2018-06-28 20:02 GMT+02:00 Maciej Jagielnicki :

...

Thank you for the suggestion.

Nothing changes when I run phenix real space refinement with SS restraints turned on. I can either input a .pdb file with correct SS information in the header or a .pdb file with no SS information at all and the end result (new refined .pdb) is the same file with the same SS information in a new header. The information stays the same and 'correct'.

The main problem in my original question is that when I (or the pdb extract tool that converts a .pdb file to a .cif file for submission to PDB) remove the header SS information altogether (why you would do that for PDB deposition is unknown to me), that resulting .pdb file still shows SS elements in different pdb viewers but SS info is no longer correct. How can you have SS elements displayed if it description is missing from the header? Maciej

On Thu, Jun 28, 2018 at 12:38 PM, Pavel Afonine wrote:

...

Hi Maciej,

please have a look at item #10 and Notes section here:

http://www.phenix-online.org/documentation/reference/real_sp ace_refine.html

One sentence summary is: Make sure you use secondary structure (SS) restraints in refinement, also making sure you provide SS annotations as HELIX/SHEET records in input PDB that are as accurate as possible; this is critical for successful refinement.

If you have more questions please get back to us!

Good luck! Pavel

On 6/28/18 09:23, Maciej Jagielnicki wrote:

Hi All,

I used a combination of phenix real space refine and Coot to build a model into a cryoEM density map. I used an existing .pdb file as a starting point for my building. During the build, some helical elements became strands and vice versa (it was very clear to my eyes). Unfortunately, neither Phenix nor Coot would automatically update the header information for secondary structure elements, and I had to manually edit the header so that helices would show up in correct places when the .pdb file was viewed in Chimera or similar software. Not an ideal way of doing this which is likely related to what happened next.

When the header information is removed, however, as it happens when the pdb.extract tool prepares a .pdb file for submission by turning it into a .cif file, the secondary structure elements are still displayed when a .cif file is visualized (bizzare in itself) but also the secondary structure is taken straight from the starting .pdb file (the one I used as a base for my model to which there is no connection whatsoever). The question is this - where does the incorrect secondary information come from in the absence of the header? How to fix this? I cannot deposit a file that shows helices/sheets wherever it wants...

Regards, Maciej J.

-- Maciej Jagielnicki, MS. Graduate Student University of Virginia Molecular Physiology and Biological Physics 480 Ray C Hunt Drive Rm 320 Charlottesville, VA Lab Phone: 434-243-2520

-- Maciej Jagielnicki, MS. Graduate Student University of Virginia Molecular Physiology and Biological Physics 480 Ray C Hunt Drive Rm 320 Charlottesville, VA Lab Phone: 434-243-2520

_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]

-- Maciej Jagielnicki, MS. Graduate Student University of Virginia Molecular Physiology and Biological Physics 480 Ray C Hunt Drive Rm 320 Charlottesville, VA Lab Phone: 434-243-2520