You are here

Connecting 2 Domains Via Non-Flexible Linker

3 posts / 0 new
Last post
Connecting 2 Domains Via Non-Flexible Linker


I originally posted my question regarding modeling of a multi-domain oligomer under a different thread. After getting a lot of useful answers and consulting Ingemar Andre off the list I am still facing one major problem.

The protein at hand consists of two domains A and B connected by a structured linker. Structural models for all three "parts" are available. How would one connect these to obtain a full-length model?

This refers specifically to Ingemar's and Rocco's following advise for obtaining a oligomeric structure of my protein:

1) Make the domainA-linker-domainB model(s).
2) Symmetrically dock the model(s)
1) Separately dock domainA and domainB
2) From a set of decent looking models from self-docking with domain A and domain B find combinations of dimers of A and B that would be able to connect with a linker.
3) Place the domains in such a manner that a linker can connect A to B.
4) Connect the linker to the proteins using loop modeling to close the covalent links between A, linker and B.

Thanks once more in advance.

Post Situation: 
Mon, 2011-09-05 06:10

I think I solved the problem. The assembly protocol and standard loop modeling either did not work at all or not satisfyingly.

A combination of different forum posts and ideas finally helped me to come up with the following protocol which works like a charm in my case. Comparative modeling is the key. The following protocol is a good way to start.

What you have and what you want to do:

You have structures (pdb files) of individual domains of a target protein and want a model of the full-length protein.

I use the term "target protein" to refer to the full-length protein you want to model. "Template" refers to the input pdb files of individual domains.

What you need:
- a fasta file of your target protein
- fragment files of your target protein
- alignment file
- template pdb file --> pdb file containing the structures of all individual domains you have as template
(-optional: constraints and/or symmetry definition file)

The alignment file looks in general as described in this post to the forum. In this case the target line is the sequence of the full-length protein you want to model. Template is a pseudo-alignment of the amino acids of all the domains you have with respect to the full-length protein. The following example might help:


The following protocol runs with the minirosetta application.
-run:protocol threading
-in:file:fasta target.fasta
-loops:frag_files ./fragments/target_09_06.200_v1_3 ./fragments/target_03_06.200_v1_3 none
-loops:frag_sizes 9 3 1
-loops:remodel quick_ccd
-loops:relax fastrelax
-relax:minirelax_repeats 8
-score:weights score12_full
-in:file:alignment alignment.aln
-in:file:template_pdb template.pdb
-database /opt/rosetta-3.2/rosetta_database
-nstruct 1000
-out:user_tag default
-out:file:silent default.out
-out:file:scorefile default.scr

optional for constraints:

-cst_file ./constraints/110805_ytva_constraints.wofmn.cen_cst
-cst_fa_file ./constraints/110805_ytva_constraints.wofmn.fa_cst
-cst_weight 1.0
-cst_fa_weight 1.0

optional for symmetry:

-symmetry:symmetry_definition target_symdef

Fri, 2011-09-09 01:38

hi, thanks for sharing your idea! Could you give an example how you included your domain coordinates in your "template.pdb"?

Tue, 2015-07-14 04:30