Hey there,
I'm having trouble getting erraser to run. The following command:
erraser.py -pdb model_real_space_refined.pdb -map ../maps/FFT_13/70S_converted_1.ccp4 -map_reso 3.73
Give the errors as seen in the attached output.txt.
Everything compiled fine, using gcc5.4
any help would be appreciated.
cheers
matt B
Attachment | Size |
---|---|
![]() | 5.62 KB |
Category:
Post Situation:
I'm pretty sure this is a bug with the RNA code using the wrong initialization calls (core init instead of protocols init or devel init). The first thing to try is to use a recent weekly release and see if that works. If not we'll try to fix THAT one instead of trying to fix 3.7.
Similar issues with the lastest build as well. See output below:
Also here is my .bashrc, as it appears in this instance its a path issue?
erraser.py output:
Ok.... So I fixed the launching problem by going into:
$ROSETTA/main/source/bin and just making a symbolic link to erraser_minimizer
Seems to at least start now. Seems like a bug in erraser_util.py
Still not working:
So went and made another symlink for resample_full_model
Then these errors occured (see attached txt file for output).
Is there something critical that I have missed in setting things up?
No, not really.
The lab that maintains ERRASER uses a slightly different build system than the standard Scons build system. They've customized this build system such that it also produces non-labled executables. (So they get a `main/source/bin/resample_full_model` in addtion to the `main/source/bin/resample_full_model.linuxgccrelease` or the `main/source/bin/resample_full_model.macclangrelease` executables.) When they set up the erraser running scripts, they apparently forgot that not everyone is using that build system, and so are using the unlabeled executable names, rather than the labeled ones that the Scons build system produces.
Doing the symlinking as you're doing is a reasonable way of handling the issue.
Hey, so if this is the case that this should be built in a different way, is there some documentation I can follow? Because I'm a bit stuck trying to fix the error below. Seems like the python script is trying to find something that isn't there? A module called, parse_tag.
Try going to main/source/src/apps/public/ERRASER/erraser_util.py line 52 or so, and add a dot to the beginning of "linuxgccrelease", like it is with linuxclangrelease, macosgccrelease and macosclangrelease.
So adding the dot did help to the erraser_util.py script. But now this error.....
So then I put the following in my environment variables, as it clearly wanted the right pdb2fasta.py
and then got this error:
Now, i'm completely stuck again.....
Can you check the 'seq_rebuild.out' and 'seq_rebuild.err' files to see what error messages are being printed there?
See below for the log outputs. Maybe something wrong with the PDB file?
seq_rebuild.err
seq_rebuilt.out
To follow up on this, I think the problem is still the pdb2fasta.py as the fasta file in the rebuild directory is just full of XXXXXXXX with no sequence data. So perhaps I need to point to a different pdb2fasta.py in my environment variables?
There's an 'INSTALL' script in the rna_tools subdirectory tools repo. Try typing `source tools/rna_tools/INSTALL` (and consider adding it to your .bashrc). That should fix your PYTHONPATH so it finds parse_tag, and pdb2fasta is more likely to work.
The hacky way is to manually edit that fasta file so that it says what it 'should say' but that is likely to help you for one step of the protocol and then fall apart a little thereafter.
Hey so I sourced that INSTALL script.
Similar problem:
Maybe something is up with the version of Phenix installed?
Maybe. Hard to say without more than just the backtrace, but maybe it isn't finding resample_full_model or swa_rna_main the same way it wasn't finding erraser_minimizer.
I can try to run it myself, if you would feel comfortable sharing the data (privately; I can give you my email); that might help debug faster.
Sounds great! Please email me on matthew.belousoff@monash.edu
I'm pretty sure its finding resample_full_model, and the issue lies somewhere in the pdb2fasta.py as it still spits out a meaningless fasta file (full of X instead of residue identifiers).
cheers!
matt B