latest version of rosetta–unable to build models

[Anonymous]

hi,

I have been successfully following the protocol as per the paper “Small-molecule ligand docking into comparative models with Rosetta” to build the homology models of transporters using Rosetta 3.5. However, recently, after installing the latest version of rosetta :rosetta_2014.35.57232_bundle and tried to build the models using the protocol as mentioned in the paper (mentioned above) I could not build the models. The program is running without errors but could not fold the protein. I was wondering if i have to follow a different protocol for the latest version of rosetta? and how this issue could be fixed?

Please let me know your suggestions.

Thanks,
Pramod

[Anonymous]

At what stage are you seeing the issue? When you say “could not build the models/could not fold the protein” how does that show itself? Are you getting any error messages? Are you getting any output structures? If so, what do the output structures look like? Are there any files you expected that you don’t get – or any that you don’t expect and do get?

Off hand, I’m not aware of any changes which would dramatically affect the protocol, but that doesn’t mean there aren’t any – a bit more information will help to track things down.

[Anonymous]

hi rmoretti,
my bad–i have models as outputs with scores, but they are not folded . They donot look like template in the core structure. Modelled protein has helices but they are not arranged as the template should, not even close, they are scattered and randomly arranged. I am not getting any error messages.

Please let me know any other information or files that I can provide to resolve the issue?

Thanks,
Pramod

[Anonymous]

My initial thought is that you likely have an issue with your alignment. If the alignment is off – or worse, contains the incorrect number of residues – then your structure will fall apart when you try to model it. Make sure when you do your alignments you’re taking the sequence you want to model, and aligning it to the sequence that’s actually in the PDB. Note that in many cases this is different from the fasta sequence given by the RCSB. The latter is the crystallized sequence, and the PDB sequence might have gaps due to unresolved/unstructured regions. Instead, get the fasta for the template from the clean_pdb script, or from the script main/source/src/python/apps/public/pdb2fasta.py run over the cleaned template PDB.

what I am doing: I am using clean_pdb script…

It’s good to take a look at the threaded PDB in PyMol or the like, and check that the sequence lines up like you want it to, and that the inserted residue are in the appropriate (sequence) location. (The inserted residues will have all-zeros coordinates in the threaded PDB – so they won’t be placed in 3D, but should be present in the appropriate location in the amino acid sequence.)

What I am doing: I am wondering why all residues in my threaded.pdb has zero coordinates

Also, makes sure that all your input files (fragments, loops, etc.) have been generated with reference from the same target fasta. If you have numbering mismatches there, you can get bad structures out due to the mismatches.

What I am doing: I have given the correct fasta file to generate the fragments and loops

[Anonymous]

my bad again: no my threaded.pdb has zero coordinates for resisdues for inserted resides only

[Anonymous]

My initial thought is that you likely have an issue with your alignment. If the alignment is off – or worse, contains the incorrect number of residues – then your structure will fall apart when you try to model it. Make sure when you do your alignments you’re taking the sequence you want to model, and aligning it to the sequence that’s actually in the PDB. Note that in many cases this is different from the fasta sequence given by the RCSB. The latter is the crystallized sequence, and the PDB sequence might have gaps due to unresolved/unstructured regions. Instead, get the fasta for the template from the clean_pdb script, or from the script main/source/src/python/apps/public/pdb2fasta.py run over the cleaned template PDB.

It’s good to take a look at the threaded PDB in PyMol or the like, and check that the sequence lines up like you want it to, and that the inserted residue are in the appropriate (sequence) location. (The inserted residues will have all-zeros coordinates in the threaded PDB – so they won’t be placed in 3D, but should be present in the appropriate location in the amino acid sequence.)

Also, makes sure that all your input files (fragments, loops, etc.) have been generated with reference from the same target fasta. If you have numbering mismatches there, you can get bad structures out due to the mismatches.

[Anonymous]

Can you give us the full command line you are using to generate these unfoldable models? Attach any flags files you used to generate them as well? How did you do the alignment and threading? Have you used the exact same inputs/sequences on both versions or Rosetta?

[Anonymous]

That’s how it’s supposed to be. You’ll get zero coordinated for the unaligned residues, and non-zero coordinates for the backbones of the aligned residues.

When you run the loop remodel step, be sure to list all the unaligned regions as loops in your loops file, and Rosetta will rebuild the backbones for those missing regions, and assign coordinates for the missing atoms.

[Anonymous]

I apologize for the delayed response.
This the command line I am using for loop modeling after generating the threaded.pdb

/home/henrylab/Applications/rosetta-3.5/rosetta_source/bin/loopmodel.linuxgccdebug -database /home/henrylab/Applications/rosetta-3.5/rosetta_database -loops:remodel quick_ccd -loops:refine refine_kic -in:file:s 2ou0_threaded.pdb -in:file:fullatom -loops:loop_file 2ou0_.loops -loops:frag_files aat000_09_05.200_v1_3 aat000_03_05.200_v1_3 none -nstruct 2 -ex1 -ex2 -loops:extended true -loops:idealize_after_loop_close -loops:relax fastrelax -loops:fast -out:file:silent /home/henrylab/Desktop/rdat4M48rosetta/silent.out

I used clustalw for alignment
command line for generating threaded.pdb

python /home/henrylab/Applications/rosetta-3.5/rosetta_tools/protein_tools/scripts/thread_pdb_from_alignment.py –template=4M48_A –target=2OU0_ –chain=A –align_format=clustal alignment.aln 4M48_A.pdb 2OU0_threaded.pdb
I cannot attach the flag files as the site deos not allow it for its size

Yeah I have used the same inputs and sequences on both versions

[Anonymous]

/home/henrylab/Applications/rosetta-3.5/rosetta_source/bin/loopmodel.linuxgccdebug -database /home/henrylab/Applications/rosetta-3.5/rosetta_database -loops:remodel quick_ccd -loops:refine refine_kic -in:file:s 2ou0_threaded.pdb -in:file:fullatom -loops:loop_file 2ou0_.loops -loops:frag_files aat000_09_05.200_v1_3 aat000_03_05.200_v1_3 none -nstruct 2 -ex1 -ex2 -loops:extended true -loops:idealize_after_loop_close -loops:relax fastrelax -loops:fast -out:file:silent /home/henrylab/Desktop/rdat4M48rosetta/silent.out

I used clustalw for alignment
command line for generating threaded.pdb

python /home/henrylab/Applications/rosetta-3.5/rosetta_tools/protein_tools/scripts/thread_pdb_from_alignment.py –template=4M48_A –target=2OU0_ –chain=A –align_format=clustal alignment.aln 4M48_A.pdb 2OU0_threaded.pdb
I cannot attach the flag files as the site deos not allow it for its size

Yeah I have used the same inputs and sequences on both versions

looking forward for suggestions

[Anonymous]

So are you getting the same as before, with the scattered helices and such? Which part is giving you the trouble, the threading or the loop modeling? You should be able to attach your flags file – they should be short files, but it seems you have the command line here. Can you attach the output PDB that is funky, the post-threaded model, and the starting template as txt files? Have you tried running one of the weekly releases?

[Author]

Posts