I want to run a relaxation on a protein structure that I've got already. It's a membrane protein so I wanted to be able to use the relax application but with the membrane scoring functions and scoring weights from the membrane ab initio application.
I've tried running relax mode, specifying the scoring weights to be 'score_membrane.wts'; the command line I used is shown below:
''relax.linuxgccrelease -database /usr/dunn/software/rosetta/rosetta_database/ -relax:fast -score:weights score_membrane -out:pdb -out:prefix fast_memb_test -out:nstruct 1 -in:file:s input_files/1OKC.clean.pdb''
but this generated a segmentation fault after the line:
''protocols.relax.SimpleMultiRelax: Running FASTRELAX fast_memb_test1OKC.clean_0001''
I had a look at the apply function in SimpleMultiRelax - one of the things that is done just after that ouput line is generated is that the original pose is scored, so this is most likely the problem, but I don't know what to do about it.
First of all, is it even possible to use the membrane scoring functions in this context and then, if so, what inputs should I give?
I'm sort of surprised that this doesn't work. Can you repeat it with the debug executeable? Most of the segfaults are caught and give error messages when using debug mode. You'll have to recompile without mode=release (use nothing, or mode=debug instead).
I think this ought to work so I'm not sure why it isn't.
Yes, membrane relax will not work in Rosetta3.1 - Patrick Barth (developer of membrane relax code in Rosetta) recently added it to the main Rosetta code, so I think it will be available in the next Rosetta release.
Ok, thanks for letting me know.
I did try using the debug executable before I saw the latest message and it seems like the problem was to do with a public function in the DataCache class, 'get_const_ptr'. I don't understand exactly what the problem is but the function is called in core/scoring/MembraneTopology.cc and core/scoring/MembranePotential.cc, so I can see why things don't work with the membrane weights, but I'm still not really sure how to fix it.
Do you know, on an approximate time scale, when the next release of Rosetta will be?
I think we're looking at September or October for 3.2. The Baker lab devotes a lot of resources to CASP. We discussed this a few weeks ago and there wasn't much developer interest in an April release before CASP.
The DataCache is basically a "bag of holding" used to pass pose-related stuff around the code attached to a pose. Probably the Membrane score function is expecting certain data objects inside (probably scoring-related neighbor graphs which tell it what residues to score how), but nothing is actually setting the data up, thus the crash. You may be able to find the code that sets it up in the membrane_abinitio code and patch it into relax.
You also could consider trying the membrane abinitio/relax mode with fragments drawn from your pre-existing known structure?
Not to resurrect a dead horse, but is drawing fragments from a the "pre-existing know" model then running membrane_abinitio still the solution to this now that rosetta 3.3 is out? I've been searching for something like membrane_relax, but if it exists, I'm having a hard time finding it in the documentation.
Relatedly, in case there is a membrane_relax, would just the membrane scoring functions be sufficient for this to be accurate or would the .span and .lip4 files need to be involved too? I assume they'd be used in the case of the membrane_abinitio/fragments-from-model solution.
Thanks from a rosetta noob.
Thanks to Patrick Barth and the Meiler Lab's tutorials, I was able to construct the attached input file for relaxing a membrane protein. This is run with the following commandline:
relax.linuxgccrelease -suffix _$i -seed_offset $i -out:file:silent 2rh1_1xio_$i.out -out:file:scorefile Menv_smooth_score_$i.sc @Menv_smooth_Tl.txt &
Note that I have my ROSETTA3_DB environment variable set. I use a pbs queuing system, and the &i iterates over the number of nodes requested in the submission script.
Also attached is my spanfile as an example. This forum doesn't except .span as an extension for attachments so I've tacked a .txt onto it.