Hi. I am trying to fix atoms (or residues) in place during the relax application. After searching through the forum, I saw that MoveMap could be used to accomplish this for the FastRelax application. I assume this is the same thing as the "fast" option for relax in the newest version of Rosetta. The old post I am referring to can be found here http://www.rosettacommons.org/content/frelax-relaxation-constrains. However, I am unsure how the commands listed at the end of the post can be applied. Can they be used within the options file? If so, how would one "set up an instance" of MoveMap within the relax application? Thanks for your help.
Matt
Post Situation:
Are you trying to fix residues in place in 3-D coordinates (relative to an arbitrary origin), or in internal coordinates (no motion relative to their neighbors)? If you fix internal coordinates for a smattering of disconnected residues, they will end up in altered 3-D coordinate locations due to the fact that Rosetta refolds through the ICM AtomTree.
If you want internal coordinates fixed (which allows absolute coordinate fixation so long as the AtomTree root (read: first residue) is included in a contiguous fixed portion, try passing a user-defined MoveMap to FastRelax within RosettaScripts as described here: http://www.rosettacommons.org/content/relax-resfile.
If you just want residues in general not to wander far from the crystallographic starting point, use the option (I'm not sure of the spelling) constrain_relax_to_start_coords. This won't ensure 100% lack of movement but it's pretty good. I don't know that there's a way (from commandline, without writing your own constraints file) to have part of the protein be restrained in this way and part not.
"I am unsure how the commands listed at the end of the post can be applied"
Which exact ones did you mean?
Hi. I need the absolute coordinates set in place, and is there any way possible to ensure that they do not move at all (that they are absolutely fixed)? The commands I was referring to were ".init_from_file" and ".show", but if MoveMap cannot keep the 3D coordinates locked in place, then it is not necessary to explain these commands. Thanks again for all of your help!
I am not aware of a way to do what you want from command line.
Let me explain why this seemingly simple thing is hard. Rosetta folds proteins using internal coordinates (lengths, angles, torsions), starting from the root of the AtomTree (usually the first residue's N). Any torsional change propagates to move things downstream. Thus, and torsion change between the root and the atoms you want to not move will cause them to move with respect to the stuff before the moving torsion and with respect to your fixed 3-D coordinate frame.
If you wanted residues contiguous with the atom tree root fixed, we could probably do that, but if you want residues scattered through the protein fixed, that's much harder.
It is possible to set up an AtomTree (FoldTree, really) such that all your nonmoving positions are connected together and won't move as a unit compared to your reference frame, while the other positions relax around them. The problem here is that it requires breaking up the internal coordinate framework of the protein, which introduces artificial "chainbreaks" where residue N+1 is not kinematically dependent on residue N. The problem here is that the beautiful peptide bond connection between N and N+1 is maintained only imperfectly if they aren't actually folding together - Rosetta knows what that bond ought to look like, but it's a mobile degree of freedom lumped in with all the rest, so it gets moved some anyway. The easiest way for you to do this is probably PyRosetta; you'd set up a custom FoldTree (which makes an AtomTree) and a MoveMap, then you'd get the fixed positions you want.
It's worth considering _why_ you want some atoms to be totally fixed in the 3-D coordinate space. Most applications are tolerant of some relaxation. If you trust your inputs to be 100% correct for those coordinates, why relax the rest at all? If it's enzyme design, I'm pretty sure that they allow the theozyme to become imperfect for the sake of threading it onto a real protein backbone, rather than try to hold the theozyme and try to repair the backbone (I can check, but it'll be in the papers anyway)
I agree with Steven. It would be a very rare situation where you would need to keep the coordinates absolutely fixed. Despite how conventional presentation would lead you to believe, there can be quite a bit of experimental error in crystal structure coordinates. With coordinate constrained relax (http://www.rosettacommons.org/manuals/archive/rosetta3.4_user_guide/d9/d...) and moderate constraint weights, you can typically keep the atomic coordinates within 0.1 Ang of their starting location, which (depending on resolution) can be within experimental error.
I can confirm Steven's recollection that in enzyme design (where we have theozyme coordinates from quantum mechanical calculations) we don't typically fix the coordinates absolutely. Doing so would indeed make it difficult to find a placement onto a existing scaffold. Instead, we typically allow the theozyme geometry to move, but impose constraints on it to keep it from deviating from the ideal structure.
That would be my suggestion on how to proceed. If you relax you requirement for absolute coordinate accuracy, you can instead impose coordinate constraints on those atoms which you wish to fix the location of. If you increase the strength of the constraints enough, you'll be able to keep them practically fixed within a very small distance of the desired location, even if it isn't theoretically fixed.
Thank you both very much for your help. I will use strong coordinate constraints to maintain the geometry of the theozyme.