You are here

The Usage of motifgraft

3 posts / 0 new
Last post
The Usage of motifgraft

Motifgraft is a is a rosetta implementation which grafts a motif from a source structre into new scaffolds. In order to use motifgraft, three input structures are needed:

  1. Motif, which is the fragment (or fragments) that the user want to graft.
  2. Context, which is the protein (or proteins) that interact with the motif.
  3. Pose (or list of poses), that is the scaffold in which the mover will try to insert the Motif. 

I am now trying to use motifgraft to graft a fragment from a complex structure which contains three chains (chain A, H and L) to a set of new scaffolds (around 400 structures). According to the tutorial of motifgraft (, a XML code for input looks like:

<MotifGraft name="motif_grafting" context_structure="./contextStructure.pdb" motif_structure="./motif_2NM1B.pdb"  RMSD_tolerance="1.0" NC_points_RMSD_tolerance="1.0" clash_score_cutoff="5" full_motif_bb_alignment="1" revert_graft_to_native_sequence="1" allow_repeat_same_graft_output="0" />

      I select 20 residues in chain H in my source complex structure, save it as a seperate pdb file, and use this file as the input for the option "motif_structure"; Because this motif interact with chain A, I save chain A as a seperate pdb file and use it as the input for the option "context_structure" . Are these input files correct for the MotifGraft Mover?

     What I cannot understand now is that these is no option for the input of Pose (scaffold) in the above example xml code. The pose in my situation is a set of structures which I save in a directory named "pose". How should I provide this pose information to the MotifGraft Mover?

Another question is that which rosetta command should be used to call the MotifGraft Mover. I guess it is "rosetta_scripts.default.linuxgccrelease". The name of my input xml code file is "motifgraft.xml", and I guess I should write the command as the following: 

      $ROSETTA3/bin/rosetta_scripts.default.linuxgccrelease -parser:protocol motifgraft.xml

Am I correct?

Thank you in advance



Post Situation: 
Sun, 2017-11-19 20:22

A more complete tutorial for MotifGraft can be found at Under "Epitope Grafting".

Regarding your command, you're correct that you'll use the rosetta_scripts executable. Like all usage of rosetta_scripts, though, you won't just give the XML. You'll also need to provide the starting structure for the run (along with any other input/output controls). That's where your scaffolds are being provided to MotifGraft: through the -in:file:s option passed to rosetta_scripts. Generally speaking, if you're interested in a number of potential scaffolds, you'll attempt to graft the motif onto a number of different input scaffolds, and then post-filter any of the output structures. Some will be good, but a fair number may give you rather bad structures, or potentially no structures at all (if they fail to graft completely).

Mon, 2017-11-20 07:52

Hi, rmoretti

Thank you for your reply. Now I have edited the MotifGraft mover as this:

<MotifGraft name="motif_grafting" context_structure="../context.pdb" motif_structure="../motif.pdb" RMSD_tolerance="1.0" NC_points_RMSD_tolerance="0.25" clash_score_cutoff="5"     full_motif_bb_alignment="1" revert_graft_to_native_sequence="1" allow_repeat_same_graft_output="0" />

and run the xml script using:

cd pose
    for file in  $files; do rosetta_scripts.default.linuxgccrelease -s $file -ignore_unrecognized_res -parser:protocol multigraft.xml -out:path:pdb ../outputs -out:path:score ../outputs; done

This generates many pdb files and a file in the oupts directory. Each of the genereated pdb files corresponds to its original scaffold structure. I expected that the context structure (chain A) would be incorporated into the generated structures, but it didn't.    I have checked some of the pdb files and find none of them contains the motif to be grafted. Does it mean that they do not match the criteria for graft? Should the context structure be incorporated into the generated structure? How can I know  whether graft occurs for each of the generated structure?

I am also reading the tutorial at Meiler lab webpage. It says that MotifGraft only works with monomeric scaffolds as grafting targets. My scadffolds are all antibodies that contain two chains (H & L). Does that mean that MotifGraft will never work for my situation?

Best regards.

Mon, 2017-11-20 19:20