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.
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:
target 1 MITAKLSERQWPAYCVTSIKAEDEEKKGFQNAPILGWFQMANDDYCTIG
template 1 ---------QWPAYCVTSI---------FQNAPILGWFQMANDDYCTIG
The following protocol runs with the minirosetta application.
-loops:frag_files ./fragments/target_09_06.200_v1_3 ./fragments/target_03_06.200_v1_3 none
-loops:frag_sizes 9 3 1
optional for constraints:
optional for symmetry:
hi, thanks for sharing your idea! Could you give an example how you included your domain coordinates in your "template.pdb"?