I'm currently working on generating a DNA complexed protein and I thought I could use the relax application to avoid steric clashes generated during the "copy/paste" approach I used to put the DNA and protein together (after alignment with previously solved structure).
Does anybody have experience with improving the quality of DNA-protein complexes through Rosetta? my aim is to prepare the PDB for conventional molecular dynamics.
Any help is appreciated.
Relax should indeed be able to handle protein-DNA complexes. The one caveat is that you may need to tweak the nomenclature of the residues and atom names in the DNA portions. I think Rosetta should handle the standard wwPDB conventions, but there may be some manual adjustments you need to do, especially if you're going back and forth to a MD program, which may have its own ideas about things.
If you were doing heavy modeling of DNA, there are certain tweaks to the settings you might want to fiddle with, but for a general "make sure things aren't clashing", the standard ref2015 energy function should be okay. If you've just jammed two structures together, I'd probably recommend using the -constrain_relax_to_start_coords option for relax to keep things from falling apart due to clashes. (I probably wouldn't recommend -ramp_conastraints false or -coord_constrain_sidechains, given that it's not an actual experimental structure.)
Keep in mind that the energy function used by Rosetta and the energy function used by your MD program are slightly different, so relaxing with Rosetta isn't going to (entirely) replace the equilibration period you need before your MD run. But it probably will indeed help to get over the major issues with manually putting together the complex, and the small clashes/imperfections that naturally arise from that.
Thank you for your kind answer. Those were exactly my plans, so that gives me confidence about my workflow.