Mutations are ignored (fixbb) when using -auto_setup_metals option

[Anonymous]

When using fixbb.static.linuxgccrelease (2018 or 2019 release) with the -auto_setup_metals option set, see cmd line below, all mutations to a residue with a link to a metal ion, are ignored in the model generated (PDB file).

Run process: /usr/local/rosetta/main/source/bin/fixbb.static.linuxgccrelease -s ../src_updated_0001.pdb -extra_res_fa ../sub.params -resfile resfile-S100.txt -nstruct 1 -ex1 -ex2 -database /usr/local/rosetta/main/database -jran 1001 -constrain_relax_to_start_coords -use_input_sc -minimize_sidechains -overwrite -auto_setup_metals

Maybe we are misunderstanding how it’s intended to work but we were under the impression that you would need to add the -prevent_auto_setup_metal_bb_variants for that to happen. Just to make sure, we tried it with and without that option and the results were the same: the mutations to the residues with the links were ignored. If we don’t use -auto_setup_metals then it works as expected.

Is there anything we can do for the mutations not to be ignored when using -auto_setup_metals? Is this by design or is it unintended (or maybe we are missing the obvious)?

Thanks,

Hector

[Anonymous]

prevent_auto_setup_metal_bb_variants interacts with metals bound to backbone heavy atoms.  You can still design in that case because the backbone does not move during design.

 

I don’t think you can design metal binding sites with auto setup metals.  The whole point of the method is to tell rosetta a metal binding site is present so it knows not to move it around too much; designing it away is the opposite of that.  If you want to get rid of it, just don’t use the flag.  If you want to have a metal site but with different residues in it…I don’t think there is an “easy mode” for that, I think you need to generate your geometries manually.  Note that stock Rosetta’s scorefunction does not “know” what a metal binding site looks like (tetrahedral vs octahedral, etc) – the point of auto_setup_metals is to encode the starting geometry as “correct” via constraints.  Following from that, it won’t know what “looks right” for a site altered to have a different set of liganding atoms.

[Author]

Posts