Hi Tommi,
Yes, mr_rosetta requires that the sequence in your model matches the sequence file (though there can be gaps). After autobuild there may be some segments (as in your case) that are not assigned to sequence and that may have whatever sequence matched the map. This will indeed cause trouble with mr_rosetta.
Probably the best solution is to remove all the segments in your autobuild model that are labeled "UNK", indicating that the sequence is not known.
You could also try running:
phenix.assign_sequenc?e
on your model/map/seq file to see if you can get any more of the model assigned to sequence.
I'm not quite sure what the command line vs command file problem is, but if it causes problems for you I'm happy to have a look if you send everything I need to check it out,
All the best,
Tom
________________________________
From: [email protected] on behalf of Kajander, Tommi A
Sent: Wednesday, December 21, 2016 7:18 AM
To: [email protected]
Subject: [phenixbb] alingments in mr_rosetta
Hi,
I am wondering about the alignment of sequence and PDB model in mr_rosetta - the program seems to complain if the sequences dont
exactyl match - so is this really the case? (I have an initial model from autobuild with some fragments that do not have fully correct sequence - hence
it doenst exactly match with the original sequence of the protein). Does it have to? Or should there be a way around it?
also the command file examples online e.g. :
phenix.mr_rosetta \
seq_file=seq.dat \
data=coords1.mtz \
search_models=coords1.pdb \
already_placed=True \
rescore_mr.relax=False \
rosetta_models=20 \
ncs_copies=2 \
space_group=p212121 \
use_all_plausible_sg=False \
nproc=200 \
group_run_command=qsub
when using the command file all the commands are not recognized for some reason (running on a mac with bash shell) e.g.
rescore_mr.relax=False, same with fragment files... not quite sure what the problems is there, probably something trivial.
-if i just enter them on the command line it works.
Thanks,
Tommi
