Angela Ahrendt , PDB files that do not have a chain label simply have an empty string as chain label, you can select those atoms that do not have a chain label by

select sele, chain ''

(two single quotes denoting an empty string)

You can use the alter function in PyMOL to change the chain id and other properties. In this link are some examples. https://pymolwiki.org/index.php?title=Alter&redirect=no

Thank you Luis Macaya I will play around with this, but there is no chain identifier in the PDB file currently, so I am thinking that I can't alter something that is not already there.

Luis Macaya I got it! The alter function did it, thanks!

For anyone else who might be a newbie like me, you can select the entire chain in Pymol using a select feature in the bottom right corner in Pymol.  In the command line type:  alter (sele), chain='A'

Angela Ahrendt , PDB files that do not have a chain label simply have an empty string as chain label, you can select those atoms that do not have a chain label by

select sele, chain ''

(two single quotes denoting an empty string)

Luis Macaya and Annemarie Honegger Thank you. Have either of you used RefineD? Now I am having a different problem. I have some students submitting PDB files to RefineD and getting results while other students are getting errors that it may not be a PDB file. Could it be as simple as that the filename can not have spaces in it? I feel guilty submitting jobs and using up the server time while trying to figure out what is going on.

What operating system are you using? Unix-based operating systems (Linux, OSX in the terminal) will not tolerate blanks in the file name, unless you put the filename in quotes - otherwise they will interpret the part of the filename after the blank as an additional parameter to be passed to the program, rather than as part of the filename. Therefore the program will not find the input file, and may also complain about a parameter it cannot interpret. It is better to generally replace the blank in filenames by an underscore character.

Generally, when you need help with a program, you make it easier for us willing to help if you give us as much information as you can:

- Which operating system you are on

- Which program and which version,

- What was your command and input triggering the error.

- The exact error messages you got

Do you have a web link to the program? RefineD is a difficult program to find on the web, since most hits on google either lead to dictionary definition, texts containing the words "refind" of "re-find" in completely different contexts or to the word refined in texts about structures if you refine the search by additional search terms). JMOL is almost as bad a name for a program, most hits being Joules per mol (J/mol ) or articles in various Journals of molecular xxx, abbreviated as J.mol.xxx

It is a web-based tool. Here is the citation for the paper. I emailed the author today.

Bhattacharya, D. (2019) refineD: improved protein structure refinement

using machine learning based restrained relaxation. Structural bioinformatics

Here is the website: http://watson.cse.eng.auburn.edu/refineD/

It's a very straightforward interface. You just upload a PDB file and select a refinement mode of either conservative or adventurous and submit the job.

I didn't capture the error, but it was in a queue for several hours before returning a failed error. I have another one queued and will save the error message and post it if it fails again.

I appreciate your feedback!

Attached is the error message that I got after uploading a PDB file to RefineD. It was in the Queue for several hours, and I suppose it didn't return the error until it was time to perform the job on the file. My PDB file was 3OZF, and I used Pymol to remove all non-protein atoms except for the magnesium ions.

Has anyone been able to edit a PDB file to remove a ligand and successfully upload it to RefineD? (See above, too.)

Thanks!

So, the method is based on Rosetta. I have looked at a few of the structures in the training dataset, and I have not found any multi-chain structures.

I am not sure whether refineD is set up to deal with non-protein components, it certainly does not mention so in the paper. The program seems to aim at refining theoretical models rather than experimental structures.

I would first try to submit a single chain and remove all HETATM records.

The presence of negative residue numbers in chain A and chain B might also cause a problem, there are programs that do not like that. You can fix these using the "alter" command in PyMOL:

alter (all),resi=str(int(resi)+6)

If a single protein chain runs OK, try a single chain with associated Mg ions (with or without the associated phosphate and pyrophosphate?).

If that is working, then you can try the full four chain object.

Removing solvent and the HPA ligand with PyMOL should not cause any problems, as long as you save the edited structure in pdb format:

save 3ozf-cleaned.pdb, 3ozf

Actually, it does say in the paper:

"We collect a dataset comprising of 181 non-redundant single- domain proteins, originally curated to benchmark the FG-MD refinement protocol (Zhang et al., 2011) spanning across multiple folds. We remove protein targets having significant sequence similarity with the training set. After removal of 31 such targets from the original FG-MD set, our benchmark dataset consists of 150 tar- gets, with average sequence identity of 18.5% (maximum 25.2%) between the test and training set and length varying from 64 to 222 residues. For each protein target, the starting structure for refinement is generated by I-TASSER (Roy et al., 2010; Zhang, 2007) using a multiple template based protein structure prediction approach."

This would mean single chain, single domain proteins with no cofactors.