Is cis-Glycine okay?

[Anonymous]

Yes, that is not a typo for cis-proline! I don’t have an issue just curious. I have a loop that Rosetta Remodel (all PIKAA entries) is struggling with despite the generic_aa set to G (but most likely for biological reasons), which got me thinking…

• Would the cis-isomer of any amino acid be sampled?
• If so, where can I find more info? (I could not find anything)
• If so would the omega deviation parameter be 180° (i.e. 0°) for a cis-isomer in the per aa values block at the end of a _0001 pdb file? If not, does any score term say it?

Thanks!

[Anonymous]

Few if any tools in rosetta will sample cis omega automatically.  I’ve seen the minimizer do it for really bad loops to fix other problems.

 

The omega score term treats cis and trans isoenergetically.  It just measures deviation from planarity.

 

The easiest thing to do is probably to start cis if possible.  Let me know what toolset you are using and we can brainstorm solutions.

 

Cis non proline are super rare and I wouldn’t trust Rosetta to predict them.

[Anonymous]

Interesting and thanks. I was interested from a theoretical point of view, so I don’t have a real concern. The protein that got me thinking was just a clinical protein and I was testing a null hypothesis that a in-frame deletion on the start of a loop did nothing: Relax -generic_aa G -num_trajectories 100 fails for the mutant case when 3 loop residues are altered, but closes 3/10 times when 6 are (at the expense of the end of helix) with a ∆∆G of 1,000 REU (the control wild type annoying did not converge with the starting ∆G (100 REU too many), but is just a question of trajectories). So I was expecting my null hypothesis to fail (and if it didn’t I would have done a GROMACS to show a RMSD difference). So while doing that, I was wondering whether a cis isomer of glycine would have been sampled and I could not find much on cis isomers and Rosetta.

About the improper dihedral/omega score in the score file, I suspected as much, so Rosetta will indeed not tell about cis amino acids, that’s a shame. It reveals HID (HIS_D), so I was slightly hopeful.

[Anonymous]

You can sample cis omega angles in loops using GeneralizedKIC.  Remodel is fragment-based, so if it samples cis-omegas at all, it only does so if the input fragments have examples with cis-omegas.  Cis peptide bonds preceding non-proline residues are very rare in the PDB (though I have seen a cis-omega preceding a glycine before).

Here’s the documentation for GenKIC: https://www.rosettacommons.org/docs/latest/scripting_documentation/RosettaScripts/composite_protocols/generalized_kic/GeneralizedKIC.  Note the sample_cis_peptide_bond GeneralizedKIC perturber (https://www.rosettacommons.org/docs/latest/scripting_documentation/RosettaScripts/composite_protocols/generalized_kic/GeneralizedKICperturber).  Since GenKIC is like a manual SLR, giving you lots of power and control at the expense of a somewhat steep learning curve involving lots of knobs and buttons, the tutorial introduction might also be useful:

https://www.rosettacommons.org/demos/latest/tutorials/GeneralizedKIC/generalized_kinematic_closure_1

https://www.rosettacommons.org/demos/latest/tutorials/GeneralizedKIC/generalized_kinematic_closure_2

https://www.rosettacommons.org/demos/latest/tutorials/GeneralizedKIC/generalized_kinematic_closure_3

https://www.rosettacommons.org/demos/latest/tutorials/GeneralizedKIC/generalized_kinematic_closure_4.

[Anonymous]

Thanks for the great explanation.

And thank you for the links! I am going to read up on that mover: I feel like I don’t use RosettaScripts enough.

But don’t worry. I am not going to do anything with cis isomers, I am very curious about how the topic is handled with Rosetta.

But from an actual concern point of view, if I were to unwittingly relax a crystallographic PDB with a cis-isomer of non-proline, Rosetta Relax would keep it in cis where the force field to agree, right?

[Anonymous]

(Note that there is a good physical reason that cis peptide bonds before residues other than proline are rare: the cis peptide bond puts the alpha carbons of the adjacent residues closer together than they would be if the peptide bond were trans.  The slight clash is energetically unfavourable.  With proline as the second residue, though, the slight clash exists either way: in one case, it’s the Calpha-Cdelta distance that’s too short, and in the other, it’s the Calpha-Calpha.  So where there’s close to no energetic penalty for a cis peptide bond before proline, there is one for everything else.  Do be aware of this before spending too much effort getting Rosetta to model cis peptide bonds.)

[Author]

Posts