I am trying to develop a model for a protein composed of multiple identical chains (a fibril actually). I know this might be a good job for the symmetry protocol, but I do not currently have enough data to define the protein's symmetry. All I have is a few constrains and some homology modeled regions. Is there a good protocol for this situation? Broker? Abinito? I am really unsure how to handle the multiple chain aspect, or if this can be handled. On a related note, if I have a multi chain system, how should I handle the fragment files?
Sure, I think symmetrical docking is what you want, but you may have to model the subunit and then model the fibril:
As for the fact that you have regions and then unknown regions, I'm not quite sure how to handle this. I'm not sure if Abinitio could take a partially folded structure and then fold the rest, but that would probably be the way to go. As for the multi chain, I think you would need one fragment file for one chain and use symmetry for the others.
In a private conversation with Frank Dimaio, he recommends the Fold-and-Dock application https://www.rosettacommons.org/docs/latest/application_documentation/str... although that's for ab inito folding if you don't have much in the way of initial structure. (I think that if you have a large amount of homology information, you might be better off homology modeling the monomer and then using SymDock protocol that Jared pointed out.)
Regarding helical symmetry (overview: https://www.rosettacommons.org/docs/latest/rosetta_basics/structural_con... ), you can either generate the symmetry file denovo ( https://www.rosettacommons.org/docs/latest/application_documentation/uti... ), or you can use a homolog structure and extract the symmetry from that ( https://www.rosettacommons.org/docs/latest/application_documentation/uti... )
Frank says both cases should be able to generate helical symmetries which are refinable by fold-and-dock, so you don't actually need to be precisely on in the symmetry definition. "Close enough" would count. (If you have a number of possibilities, you can do multiple runs with each.)
On the fragment files, if you're using the symmetry code, you should only need the fragment file for the monomer. When inserting fragments, the changes to the first monomer should propagate to the symmetry-related partners.
By the way, you may be interested in fiber structure determination papers by Sgourakis et al. They're not necessarily directly relevant, but you may be able to pick up relevant ideas:
http://dx.doi.org/10.1038/ncomms5976 (and a number of related ones: http://www.ncbi.nlm.nih.gov/pubmed/?term=sgorakis+NG)