At RosettaCON2011, many of the developers and users got together to write documentation for an afternoon under the beautiful, central Washington sunshine. I've put up a beta release of the tutorials that we assembled at RosettaCON2011 at
This is not the official release, we're still working on incorporating the suggestions from beta testers. If you find flaws, let me know here, and maybe I can get them fixed.
A tutorial that would help a lot of us non-experts would be a 'relax' tutorial which shows how to take a very large protein and only relax a small domain, while keeping the remaining domains fixed.
I've been struggling with this for a while, and I have seen a few post questioning this as well.
This was on the list of tutorials to make. Did you look at relax_a_large_structure?
That looks like a perfect solution.
One question that arises though, is whether or not this is adaptable to multiple chains.
I suppose for a multichain PDB (of which I want to relax portions of both chains), I can just join the chains into a single, and offset the numbering.
I'll give this script a shot! Thanks for bringing it to my attention.
It's going to depend on which parts you want to modify and what your underlying FoldTree is. I'd guess the default fold tree connects each chain to the root residue (first one in the PDB). That means you are unlikely to get unwanted rigid-body motions doing this sort of relax. (If you instead have a "linear" fold tree where each new chain connects to the end of the preceding chain, then this sort of relax will introduce rigid-body motions).
The scripter probably has an interface for specifying the fold tree, which you may be able to use to ensure the not-linear-fold-tree case.
Not to belabor this point too long, but do you (or anyone) have any idea how to specify the fold tree in the scripter? Or who I could contact to ask.
There is nothing documented on fold trees at: http://www.rosettacommons.org/manuals/archive/rosetta3.2_user_guide/Rose.... And the source code for the scripter is a bit to obfuscated for me to drill down into.
There's currently no general way to specify a fold tree through RosettaScripts.
Fold tree manipulation falls into the general category of "changing the pose", so it would be under the purview of individual Movers. There's currently an AtomTree mover, but that's primarily designed for protein-protein interaction design, so it may not be completely what you're looking for.
Sarel Fleishman says:
"I once put in place a way to read foldtrees from the commandline similar to the format of a silent file. I’m away in a conference so I can’t really look at that right now, but I would expect it to look something like:
-jd2:fold_tree < name of file >
It’s not the same as going through RS, but might be sufficient... Sarel."
I looked through the code and found "in:file:fold_tree". Unfortunately fold_tree is common so it's hard to find all the places the option is actually respected. Give it a try and see if it works.