The scripts and input files that accompany this demo can be found in the demos/public directory of the Rosetta weekly releases.


Author: Deanne Sammond RosettaCon Talk:

  • Computational design of a new protein-protein interface between Gi1 and a redesigned RGS14 GoLoco, Deanne Sammond, Dustin Bosch, Glenn Butterfoss, Mischa Machius, David Siderovski, Brian Kuhlman, Kuhlman lab, 2010, Session 3 on Wednesday August 4th

Our project is the computational design of a new high-affinity protein-protein interface. Our model system is an x-ray crystal structure of Gi1 bound to the GoLoco domain from the RGS14 protein. RGS14 GoLoco spans two domains of Gi1, with the C-terminal random coil region binding to the all-helical domain of Gi1. We removed this C-terminal portion of GoLoco, replacing the random coil with a de novo designed alpha helix. The redesigned GoLoco binds to Gi1 with a dissociation constant of 810nM, the correct binding of the newly designed GoLoco was confirmed using disruptive mutations at the Gi1:GoLoco interface, and the correctness of the computational design was assessed with by x-ray crystallography.

This protocol builds (or extends) a backbone for a peptide bound to a target protein, then designs a low-energy sequence.

Running the protocol

Important flags:

-ex1, -ex2, -exOH, -extrachi_cutoff 1 all seem to be very important for the sequence design run.

Example Rosetta Command Line:

rosetta.mactel aa input_pdb _ -s g000.pdb -loops
rosetta.mactel -design -l list_of_pdbs -tail -begin 342 -end 351 -chain_ -series bb -protein g000 -resfile g000_resfile -ex1 -ex2 -extrachi_cutoff 1 -exOH -no_his_his_pairE -tight_hb -try_both_his_tautomers 

How to generate tail designs:

This protocol uses 2 separate rosetta runs — one is centroid mode to build backbone coordinates and the other is a design run to find a low-energy sequence — and 2 additional scripts. Step-by-step instructions are below:

  1. Generate fragments - you can do this using the Robetta server. I named my fragment files aag00003_04.200_v1_3 and aag00009_04.200_v1_3.

  2. Make starting structure using and the g000.zones file. I also included a fasta file (see g000_.fasta) because I felt that setting the sequence improved the quality of the centroid models more effectively than using constraints. For example: -zonesfile g000.zones -fastafile g000_.fasta -parentpdb gpep_nat.pdb -outpdb g000.pdb

    The resulting file should look something like g000.pdb. The side-chains are removed from all sequence positions, and the region that will be redesigned is removed. NOTE: The input file has the sequence positions re-numbered in "Rosetta numbers", so the original pdb starts with sequence position 30, contains a "TER" between chain A and B, and chain B starts with sequence position 496. But my modified pdb (gpep_nat.pdb) starts with position 1, with no "TER" between chains A and B, and the numbering is sequential through B.

    input: g000.zones, g000_.fasta, gpep_nat.pdb
    output: g000.pdb

  3. Making centroid models

    • Copy fragments to working directory

    • Make a loop file to specity what residues can move. See g000.loops as an example. In the loop file, the 1st numer is the # of positions to be built, second number is sequence position to start design, with the final number being the sequence position where the design will end.

    • Make constraint file to make the designed region fall into the desired location OR direct the redesigned peptide toward the desired orientation using the starting sequence. I did the latter. In other words, part 2) above can generate a pdb file with all alanines in the region that will be designed OR the sequence can be specified with a fasta file (see above) so that big hydrophobics fall in buried regions and hydrophilics fall in solvent exposed regions, etc. An example of a constraint file is g000_.cst.

    • Copy over starting structure g000.pdb from step 2.

    • Run command line:

      rosetta.mactel aa input_pdb _ -s g000.pdb -loops 

    input: g000.loops, g000.cst (I didn't use constraints)
    output example: aag000_0001.pdb

  4. Merge centroid designs with fullatom starting structure (in this case gpep1.pdb) using merge_pdb.csh like this:

    merge_pdb.csh gpep1_nat.pdb [list of pdbfiles].

    You will need to edit merge_pdb.csh if you want to change which residues are being merged. For example, if you start the design at sequence position 342, like we do here, check your gpep_nat.pdb file (original all-atom pdb file) for the line # for the last atom in sequence position 341 and put this

    in after "head -", then check your centroid files for the first atom at

    position 342 and put this # after "tail -". This step is so that you don't have to repack all of the gpep1.pdb positions during the fullatom simulations. (NOTE: When building with centroid mode, we don't use a TER in between chains A and B. The TER needs to be added back in. Another merge file can be used for this.)

    input: gpep_nat.pdb, list_of_pdbs (example of pdbs in list - aag000_0001.pdb)
    output example: aag000_0001.m.pdb~ and with the TER added, aag000_0001.m.pdb

  5. Making Fullatom models:

    rosetta.mactel -design -l list_of_pdbs (the *.m.pdb merged files from step 4 above WITH a TER added between chains A and B) -tail -begin 342 -end 351 -chain_ -series bb -protein g000 -resfile tail.resfile -ex1 -ex2 -extrachi_cutoff 1 -exOH -no_his_his_pairE -tight_hb -try_both_his_tautomers -linmem_ig 10 -output_hbond_info -decoystats -group_uns 
    • -linmem_ig 10 is optional. I used it because I was running on a BlueGene and each node had very limited memory. -output_hbond_info, -decoystats and -group_uns are also optional. I used those so the output pdbs could be used with Ron Jacak's h-bond pymol plugin.

    • Interesting "feature" that seems to have appeared in this version - this call to Rosetta is looking for fragment files named aag000l03_05.200_v1_3 and aag000l09_05.200_v1_3, whereas the fragment files I used when building centroid models (3) were named aag00003_04.200_v1_3 and aag00009_04.200_v1_3. I just renamed the fragment files so I could get this done quickly.

    input: g000_resfile
    output example: aag000_0001.m_0001.pdb

See example of resfile (g000_resfile). In this file the sequence positions of the designed region are allowed to vary, and any neighboring sequence positions on the target protein are allowed to relax.

Rosetta Version

SVN Revision: 29304