You are here

Is this the intended behaviour?

3 posts / 0 new
Last post
Is this the intended behaviour?

Hi all,

I am a new PyRosetta user and am running into something weird.  I am playing with some tutorial examples to get my feet wet and here is a very simple question.  When I do the following,

p = pose_from_sequence("Q"*5) 


I get the the side chain carbonyl oxygen bonded to the backbone (see attached pdb).  Is this the intended behaviour?



test.pdb.txt6.57 KB
Post Situation: 
Thu, 2022-05-05 14:59

This is expected. By default Rosetta will build sidechains with really bad conformation. (This is both to highlight that these conformations are bad when examining it, as well as making it easier for any downstream optimization to fix them.)

Note that the sidechain oxygen isn't *really* bonded to the backbone. The PDB file format normally doesn't have explicit bonding information, so visualization programs like PyMol and Chimera use distance-based heuristics to guess which atoms are bonded together, and so will draw bonds between any two atoms which are "close enough. In your Pose, the atoms may be within whatever distance threshold PyMol/Chimera is using, but Rosetta has its own concept of which atoms are bonded or not and does not treat the atoms as being bonded. (You normally have to do special setup to get non-standard bonding of your structures.) Indeed, having them be too close will probably show up as a very bad (i.e. positive) score in the fa_rep term of the ScoreFunction if you score the Pose.

What you want to do (if you care about sidechain positioning) is to take your newly created Pose and run it through the packer (e.g. the PackRotamersMover) or similar sort of Mover (e.g. the FastRelax mover includes a packing step, so it will also optimize it.)  Given the bad fa_rep scores of the default conformers, any sort of sidechain optimization will cause the sidechains to change to a much more reasonable conformation for the current backbone you have.


Thu, 2022-05-05 15:13

Thanks Rocco! That cleared it up.



Fri, 2022-05-06 07:03