Problems when running Rosetta VIP with cofactors

Hi smlewis,

Regarding the first error: the relax protocol interpreted the bond between Fe and N correcly with the bonding information contained only in the *.params files, I did not use CONECT in the starting pdb for relax. That’s why I gessed it would work in VIP too. I will try your solution.

What I meant by “because I can comment the resfile line in the flag file and nothing changes.” is that I can make a new flag file leaving the “-resfile resfile.txt” line out (by adding “#” at the begining) and get the same error, as if rosetta was not taking it into account in the first place. On the contrary, if I leave out the *.params file, rosetta doesn’t recognize the heme and stops running (which was rather spected). I was not aware that the error from ResfileReader.cc was strictly related to the resfile option, still, the error is there with or whithout including the resfile flag. Does it make sense now?

Another piece of information: The “resfile” error occurs when rosetta is making some rotamers, you can see this in the attached log file.

This is my resfile.txt:

#Leave out heme and histidine 93

93 A NATRO
154 A NATRO

The intention of this is to not allow VIP to move or mutate the heme group or histidine 93, which binds it.

Benjamin

I see that in the solution that you proposed for the first error you refer to the VIP.cc file, but I only found VIP_mover.cc and VIP_report.cc in the protocols folder, in addition to that, I found the “core::io::pdb::build_pose_from_pdb_as_is…” line in the VIP_mover.cc file. I guess that making the change in VIP_mover is what you meant, right?

I can try doing the quick hack. Correct me if I’m wrong, but both changes would require to recompile VIP. If so, I would like to avoid recompiling all the applications. Can you give me some instructions to recompile VIP alone in a new folder in the /bin directory? something like /bin/vipAlt. Or should I go to the building forum for this?

Thanks,
Benjamin

**EDIT**
I would like to put the modified *.cc files in a separate folder and tell scons to build it as a new app, if that’s not too complicated.

Hi,

I applied the changes you proposed and compiled in a separate folder:

-Issue with the HIS-HEME bond: Changing the original line in vip.cc for “core::import_pose::pose_from_pdb( in_pose, option[ OptionKeys::in::file: ]().vector().front() );” didn’t fix the problem, I just got a new output line under the previous error message saying something similar:

core.conformation.Conformation: Failed to find a residue connection for residue 154 with connection point 1 //same line as before
core.import_pose.import_pose: Can’t find a chemical connection for residue 154 HEM //new line after changing vip.cc

-Avoiding HIS and HEME repack: The quick fix worked and I have some output structures to look at. I can give you a hand on finding a general solution for this problem, mainly in the testing end.

Going back to the HIS-HEME bond issue, shouldn’t I be able to keep the geometry of the HIS-HEME bond by using just the constraints file and avoiding repack of HIS and HEME? I tried this and Rosetta doesn’t seem to be taking the constraints.cst file into account, as I can see in the distorted angles and distances between HIS and HEME in the output structures I was able to get. Also, I don’t see any output regarding constraints as I did during the relax protocol (output muted during VIP run: protocols.simple_moves protocols::checkpoint protocols.relax core.scoring.etable basic.io.database core.io.pdb core.chemical protocols.packstat core.scoring.rms_util).

New issue 1: I realized that relaxing the structure before running VIP on it is unnecessary and might affect the outcome of the algorithm, so I added hydrogens via an on-line server leaving the HEME out and then added hydrogens to the separate HEME with pymol. I ran the mol_to_params.py script on the *.mol version of the HEME and got the *.params file and HEM_0001.pdb. To get the input structure I just pasted the content of HEM_0001.pdb before the “END” tag of my hydrogenated protein. With this input structure I ran Rosetta VIP with the following flag:

-database /…/rosetta_database/
#-ignore_unrecognized_res
-s nonRelaxedInput.pdb
-resfile resfile.txt
-extra_res_fa HEM.params
-cst_fa_file constraints.cst
-cp:cutoff 6.0
-sasa_calculator_probe_radius 1.0
-run:silent
-cp:print_intermediate_pdbs
-cp:local_relax
-cp:relax_mover “relax”
-cp:print_reports
-mute protocols.simple_moves protocols::checkpoint protocols.relax core.scoring.etable basic.io.database core.io.pdb core.chemical protocols.packstat core.pack

With this flag and the repack fix VIP works and I can get some structures, but I also get a lot of output from core.scoring.rms_util, always the same line, many times:

core.scoring.rms_util: WARNING: In CA_rmsd, residue range 1 to 154 requested but only 153 protein CA atoms found.

I couldn’t find the source of this warning, but I didn’t get it when using the pre-relaxed structure. As I said, VIP runs anyway, but should I worry about it?

New Issue 2: I’m getting some mutations in my final_mutant.pdb that are not listed among the accepted mutations in the program output. They do appear in the reports.txt file and, apparently, they are the last mutations of each run that are supposedly rejected but for some reason appear in the final pdb. This also happens when using the provided flag and input structure in the demo. A bug maybe?

Thanks,
Benjamin

Hi rmoretti,

I would like to make sure Rosetta is finding and using the constraints file, what part of the standard output should I take out of the “-mute” tag to see it?
Also, would you please point me in the right direction on how to manipulate the weight of the bonded interaction?

Thanks,
Benjamin

I was about to use Rosetta Holes to generate pdbs and score my input and output structures, but there seems to be no holes executable, as I would expect from the pdf available with the holes demo. I could not find a holes.cc file either. Should I add something to rosetta_source/src/apps.src.settings and recompile?

Thanks,
Benjamin

I went for the unified residue type option and tried to make the *.params file with molfile_to_params_polymer.py

First I had to move it to the folder where the rest of the public python apps are: /rosetta_source/src/apps/python/public so it would be able to import everything it needs. molfile_to_params_polymer.py is originally in the folder you suggested and running it from there gave this error:

File “molfile_to_params_polymer.py”, line 19, in
from rosetta_py.io.mdl_molfile import *
ImportError: No module named rosetta_py.io.mdl_molfile

After moving molfile_to_params_polymer.py, I ran it with my modified .mol file (1) as input, getting the following error:

File “molfile_to_params_polymer.py”, line 1995, in
sys.exit(main(sys.argv[1:]))
File “molfile_to_params_polymer.py”, line 1919, in main
m = molfiles[0]
TypeError: ‘generator’ object has no attribute ‘__getitem__’

I repeated this with a .mol file that had previously worked in with molfile_to_params.py and got the same error. I wasn’t able to figure out what’s wrong, but it looks like there is a problem when reading the input. I wonder if moving molfile_to_params_polymer.py to a different folder is the cause of this.

(1) modified *.mol file: I followed the instructions in the readme.txt file that comes with the ncaa demo. At one point it requires to add some lines starting with “M ” at the end of the *.mol file. I tried to add these lines the best I could, but it’s hard to extrapolate the instructions from the example to my ncaa, i.e.: what are exactly the atoms in the POLY_IGNORE line? part of the molecule that will be erased from the .params file?
I saw at the post you cited that additional lines can be used, like “POLY_UPPER” and “POLY_LOWER”, what are these lines for? Is there an alternative tutorial to use molfile_to_params_polymer.py different from the one in the demo folder? maybe one that is more extensive and less focused in the example?

Attached is my *.mol file.

Thanks,
Benjamin

Solved:

I copied all the files from the /rosetta_demo/…/scripts folder to the corresponding subfolders in the /rosetta_source/src/python/rosetta_py folder. Then, added the “UPPER_CONNECT” and “LOWER_CONNECT” lines to the .params file. Now rosetta keeps the structure of the his-heme complex as if it was a residue.