You are here

terminal amino acids of gap has incomplete valency

7 posts / 0 new
Last post
terminal amino acids of gap has incomplete valency
#1

Hello!

While doing protein minimization, I found that the terminal amino acids of gap (let say 80 and 110) have incomplete valency e.g. -C(O) and -NH instead of -C(O)O-/H and NH2/H+ as rosetta consider them very neighbours during renumbering step. 

I tried modifying both residues (let say 80 and 110) in different protein visualization plateforma and fed to Rosetta, but it modifies them to -C(O) and -NH. 

Could you suggest how to deal with it?

I hope I lucidly explianed my query, please let me know if I messed up my question.

 

Thanks!

malkeet

Category: 
Post Situation: 
Wed, 2017-11-29 05:20
malkeet.singh

Short answer:

By wanting the terminus charges on you are implying that you have two protein chains rather than one protein chain.  Change the chain designation in the PDB for the second chain to reflect that, and Rosetta should put the correct atoms on the termini.

Long answer:

If this is one protein 'in reality', with residues in the gap, but the model has a giant gap, then the region immediately where the gap starts and ends is definitely modeled wrong (because many, many atoms are missing) and so the question of what atoms to put on those pseudo-termini is unimportant.  If it's really two chemical entities, with no non-disulfide covalent bonds between the halves, then treating it as two logical chains in the PDB is the correct approach.

Wed, 2017-11-29 09:57
smlewis

Hello!

Thanks for your reply but I didn't get it absolutely.

My protein is a monomeric peptide of 1200 AAs which has two ginat gaps of about 35-40 AAs. I am using this protein for ddg calculation by using rosetta ddg monomer script. While doing the minimization, minimize_with_cst script (of rosetta) added hydrogen to the protein albeit the pseudo-termini are not logical (as rosetta perform renumering so it is taking the AA at the start and end of the gap as very neighbours and designing one amino acid as -NH while the other as -C(O) : a pseudo peptide bond - NH and CO are however not conncted in the final output structure). If I'm not wrong the pseudo termini AAs must be protonated such as -C(O)O- and NH3+. Don't you think considering pseudo amino bond in the protein will indeed affects/alter calculations in other protocols?

I'm wandering how to solve this hassle?

Thanks!

malkeet 

Thu, 2017-11-30 00:59
malkeet.singh

>  Don't you think considering pseudo amino bond in the protein will indeed affects/alter calculations in other protocols?

It will, but it's either irrelevant or we can control it.

If it is the case that your real protein has no gap, and the residues are missing from the model because of poor electron density, then that local region of your model is definitely wrong no matter what (many whole residues are missing!) and you should not try to run simulations that depend closely on details in that local region.  Running ddGs far away from those sites is fine; it won't be sensitive to a few atoms far away, and having a bad ramachandran score across the incorrect pseudobond is unimportant - it will be equally bad in all models and thus wash out of the comparisons.

If it is the case that your real protein has a gap - perhaps the intervening region was proteolytically cleaved - and you SHOULD have the extra terminus atoms present because they are really there in the real protein, then the solution is to change the chain lettering.

Thu, 2017-11-30 13:54
smlewis

> then the solution is to change the chain lettering.

Or add a TER line between the two residues. -- This allows you to keep the PDB numbering but tells Rosetta that there should be a chain break between the two residues, despite having the same chain letter.

Fri, 2017-12-01 12:05
rmoretti

Hello!

 

While using ddg_monomer script, I tried flag --ggd:dump_pdbs 'false' and 'true' both, but it is not producing any output pdb file?  any suggestion?

 

-ddg::dump_pdbs true # write out PDB files for the structures, one for the wildtype and one for the pointmutant for each iteration
Thu, 2017-11-30 01:41
malkeet.singh

A) is it running to completion?  Does the end of the log file look like it finished properly?

B) try setting the output flags, like "-out:path ." or similar.  I'm not sure of the precise names of the flags or if they are relevant, but it's worth a shot.

 

If neither of those things work, re-ask this as a separate thread; Rocco might know but I don't think he's reading this one.

Thu, 2017-11-30 13:55
smlewis