Hi Rosetta users,
I am experiencing trouble using RosettaMP to predict membrane protein structures. I am currently following the RosettaMP tutorial, which can be found here (https://www.rosettacommons.org/docs/latest/application_documentation/structure_prediction/membrane-abinitio).
To briefly describe what I have done:
1. generate structure fragments using Robetta server
2. generate a span file using OCTOPUS server and the script, octopus2span.pl
3. generate lipophilicity prediction file (.lips4 file) using run_lips.pl
I confirmed that I have nr database, blastpgp, and alignblast.pl script and when I run the script, run_lips.pl, it actually outputs _.blast file, _.blast.msa, and etc which seem proper. It seems there is no problem in running blastpgp. However, some warnings are followed after the blastpgp step such as
Use of uninitialized value highest_lipo_index in array element at /Users/siyoungkim/software/rosetta_src_2018.44.60488_bundle/main/source/src/apps/public/membrane_abinitio/run_lips.pl line 201.
Use of uninitialized value within resnum in array dereference at /Users/siyoungkim/software/rosetta_src_2018.44.60488_bundle/main/source/src/apps/public/membrane_abinitio/run_lips.pl line 201.
Use of uninitialized value lowest_lipo_index in array element at /Users/siyoungkim/software/rosetta_src_2018.44.60488_bundle/main/source/src/apps/public/membrane_abinitio/run_lips.pl line 202
I didn't copy the whole warnings as they look similar with the above warnings. In the .lips4 file, there is only one line printed, "Lipid exposed data: resnum mean-lipo lipophil entropy", but nothing else.
My OS is MacOSX and the version of perl is v5.18.2. Thanks for your help in advance.
Best,
Siyoung
Hi,ksy141,Have you solved the problem yet, I faced the same problem.
Hi,
I have also had problems with this process. Could I ask which version of the nr database you are using, and where you obtained the blastpgp application?
If it is of any help, I have found that the new ab initio TrRosetta deep-learning algorithm on the Robetta server (https://robetta.bakerlab.org) can be quite good at predicting membrane protein structures.
Hi,
I have also had problems with this process. Could I ask which version of the nr database you are using, and where you obtained the blastpgp application?
If it is of any help, I have found that the new ab initio TrRosetta deep-learning algorithm on the Robetta server (https://robetta.bakerlab.org) can be quite good at predicting membrane protein structures.
Has any solution to this been found? I'm having the same issue.
Hi all,
After a few hours in the perl debugger I managed to get this worked out. The issue seems to be that the server called to generate some of the files is dead. The script needed as a text file can be found here: http://tanto.bioe.uic.edu/lips/lips.txt
If you download that, save it as a .pl file and make it an executable, and then make sure the path to call perl in the first line are edited/correct. I also had to remove the -w/-d flag because calling these modes from inside another perl script was causing issues (that might be specific to my set-up).
Then change the portion of line #127 in run_lips.pl "curl -s $url" to "~WHEREVER-YOU-SAVED-THE-FILE/lips.pl".