I am doing unbound docking and i believe I have enough biological info to locate the binding interface and rough orientation. Doing the docking manually (or with the low res protocol) suggests that there's some conformational change in the interface that I would like to model. I'm guessing this is not something that I can accomplish with the old docking_protocol commandline app. I did see a flexible_bb_docking option but that appears to be undocumented? Maybe I can do this via the ensembles options?
I understand that everyone does docking using rosetta scripts nowadays but I don't have any experience with that and don't really see any published docking scripts to start with. There was a question asked a couple years ago but I don't see the attached script there any more : https://www.rosettacommons.org/node/3531
I've sent this along to a docking lab, but a brief answer: I don't think flexible interface docking is considered a solved problem with a "best practice" script. Certainly I've done a lot of flexible interface docking and never felt my protocols were perfect. You need to pick/design a protocol based on the problem at hand.
How much backbone change do you expect? If it's small, then just minimization may get you there - this is fairly easy. If it's middling, then the relax protocol within docking may work. This will be expensive but is still fairly "off the shelf".
If it's not small, and you need to *sample* the backbone, then it becomes a very major undertaking (both in terms of CPU time and determining exactly the right protocol).
Attached is one of the supplemental figures from a paper of mine that involved some docking (http://www.ncbi.nlm.nih.gov/pubmed/24463572), with a RosettaScript for extremely local docking with backbone minimization. It's certainly not global docking, but perhaps it can help you get started.
Maybe the conformational change I'm looking it is not that bad after all - it's a bit hard to say. I think I'll try the simpler options first.
I've done global docking with constraints to generate some starting poses. Now for minimization, I would do something like:
-dock_pert 3 8
Does that look right? Should I still have constraints on?
Now if the above doesn't look good enough, I can just run relax on some top N complexes from above I guess but what would be the proper way to call relax within docking_protocol?