You are here

Adding positional restraints to pure RNA system in DRRAFTER and rna_denovo

2 posts / 0 new
Last post
Adding positional restraints to pure RNA system in DRRAFTER and rna_denovo

I'm running into some issues applying the DRRAFTER instructions to model RNA loops between several distinct rigid-body helices, and am asking for advice here on how to proceed.

My goal is to rebuild the loop structures between two base-paired regions opf a stem loop. A fake secondary structure is like so:



(1) To get ROSETTA to parse, I have had to break up the double-helix part into its 5-prime and 3-prime strands so that rna_denovo will see the residues in the correct order. This results in effectively 3 pieces. Is there a better way to organise the rigid bodies, or is this correct?

Now, I have tentatively assigned the first RNA piece as the absolute_coordinates_rigid_body_structure. However, if I don't put in -no_csts DRRAFTER cannot proceed, as it complains about not having a protein component to automatically configure constraints. It will only work with -no_csts, for instance:

-residues_to_model A:1-59 -map_file <mrc> -map_reso 10 -start_struct ./concat.pdb -include_as_rigid_body_structures pieceA-5prime.pdb pieceB-stemloop.pdb pieceA-3prime.pdb -absolute_coordinates_rigid_body_structure pieceA-5prime.pdb -no_csts 

(2) Is there a work around for this or should I give a custom constraint file?

(3) In the custom constraint file given to rna_denovo by -extra_flags cst_file ./constraints.txt , when I give a CoordinateConstraint , which reference atom2 should I be looking at using? I have currently tried to pair atom1 and atom2 from different rigid bodies, but it doesn't seem to keep the strands together.


Post Situation: 
Fri, 2019-08-23 01:16

I've flagged this for DRRAFTER's original developer; her advice will be better than mine. The thing I can say very confidently is that the constraints behavior is expected. That automatic constraint generation is basically useful mostly for running DRRAFTER on rna-protein complexes -- if you don't have protein in your system, you won't be hurt at all by lacking constraints.

The reason you probably want constraints is because of this issue with having to specify the 5' and 3' piece separately -- understandably so. But we should be able to specify them in a single PDB. Can you share some more details about exactly what you are specifying to which flag? Your -initial_structures should be a single PDB file; it should contain as many of your -s structures as you can place confidently. As long as you specify everything in one file, you should be able to just put the residues in correct resnum-chain order within that file.

Finally, suppose you do want to provide a constraints file for some other purpose. In general, for CoordinateConstraints you want the reference atom to be something that doesn't move during the simulation. Basically, atom2 is giving you a concrete, fixed reference point to help define an absolute coordinate frame. atom2 being fixed gives "I want atom1 to be 'here'" meaning. But if you want to keep two bits of structure together, I might suggest AtomPairConstraints. That seems to be more the kind of interaction you're interested in. You shouldn't need constraints for this problem, but perhaps this will be useful in the future.

Fri, 2019-08-23 04:08