I have tested the new option -auto_setup_metals ( https://www.rosettacommons.org/docs/latest/Metals.html ) when relaxing an enzyme containing a Zinc ion, corrdinated by four cysteine residues. The weekly build version used for test was downloaded at the end of 2014 from Rosettacommon website.
The relax program finished successfully. However, the extract_pdb program fails to read the silent file generated by relax with -auto_setup_metals. It complains:
[ERROR] EXCN_utility_exit has been thrown from: src/core/chemical/ResidueTypeSet.cc line: 397
ERROR: unrecognized residue name 'CYSSG_connect'
In the silent file, the sequence contains cysteine residues followed by the unknown "residue" CYSSG_connect:
ANNOTATED_SEQUENCE:
...C[CYSSG_connect]...C[CYSSG_connect]...C[CYSSG_connect]...C[CYSSG_connect]....
Obviously, the annotation tells Rosetta that the cysteine residues coordinate the Zinc ion. But extract_pdbs program does not understant that and does not find a residue type CYSSG_connect in the res_types directory (I could not find any file with such a name or containing such a term).
I added -auto_setup_metals to the extract_pdb program, same error.
I added -ignore_unrecognized_res true to the command line, same error.
Of course a temp solution is to ask relax to output only pdb files. But is there any way to read the silent files? Was it because the residue type files are missing from the new weekly build? Was it because the residue type CYSSG_connect was not decomposed correctly by the extract_pdbs program?
I talked with the person who implemented the -auto_setup_metals feature. Unfortunately, the way it's currently constructed means that trying to read from silent files isn't going to work. You're stuck with PDBs for the time being if you're using the -auto_setup_metals feature.
(By the way, if you need to get the results out of the silent file, I think that changing the CYSSG_connect to just plain CYS (e.g. using something like the sed program) will work. Make the edit to the silent file and then extract to PDBs - I haven't tried this, though.)
The problem is more at the Zn ion, rather the CYS.
I remove the annotation (things within squared brackets) for CYS and leave the annotation for Zn: ......S[SER:CtermProteinFull]Z[ZNZN_connectZN_connect2ZN_connect3ZN_connect4]X
extract_pdbs complains:
[ERROR] EXCN_utility_exit has been thrown from: src/core/chemical/ResidueTypeSet.cc line: 397
ERROR: unrecognized residue name 'ZNZN_connectZN_connect2ZN_connect3ZN_connect4'
So I further simply the annotation for Zn to:
......S[SER:CtermProteinFull]Z[ZN]X
extract_pdbs complains:
[ERROR] EXCN_utility_exit has been thrown from: src/core/conformation/Conformation.cc line: 1402
ERROR: ZN doesnt have connection at ZN
Interestingly, if I use some pre-atuo_setup_metal version of Rosetta 3.5, the error message from extract_pdbs is different:
core.io.silent: ERROR: trying to index off the end of the secstruct array, idx is 72 secstruct_.size() is 71
obviously, the new version of extract_pdbs DOES try to do something to interpret ZN coordination in the silent file. Unfortunately, maybe files are missing in the database dir or the code is broken...
Hmm ... try also deleting all the lines starting with NONCANONICAL_CONNECTION - it looks like that causes Rosetta to still try to build the metal interaction, even though we now have a residue which no longer is set up to make such connections.
that did the trick ! cheers Rocco!