I am having problems using the ROSIE antibody section for comparative modelling of a Fv fragments, job number 33253 (Mab_Top).
Previous attempents have failed, and now I made a modification on the N-terminus (because this is was not fully sequenced) and the job is flaged as finished altough without any structure, as well as without reports of potencial errors.
I guess that there was any problem with the modelling since the CPU time used was so small 0.0147222222222.
Can someone give me some input about the files that I should have a look to find potencial errors?
Thanks in advance,
Thanks for letting us know about this.
The issue you ran into doesn't have anything to do with your job. There was an issue with the ROSIE backend where there were a bunch of out-of-date files still hanging around on the server. The ROSIE adminstrator has fixed the issue and re-started your job.
I am the one thanking you, as I said in the past I have been having problems with the N-terminus of the heavy chain, because the sequence that I had acess to misses the first 7 residues, so one way of circunventing this problem was by copying and pasting this initial residues from other IgG2a that were aligning well with my sequence.
For instante in the job nº 33227, i used the initial N-terminus of the heavy chain as
While in the current job nº33253, i used the initial N-terminus of the heavy chain as
Is there any need for a serine at positon 7 of the N-terminus of the heavy chain? And why would it be necessary? I have the codon cct at this position? So i can make this mutation in my model (S) and subsequently change it to the native residue (P) and do some energy minimization of this rotamer?
Thanks in advance,
If you haven't already, I'd recommend reading through https://doi.org/10.1038/nprot.2016.180 -- it's not exactly the method being used by ROSIE, but it's very similar, and should discuss some of the caveats and limitations.
I'd also recommend the following papers about critical determinants of antibody structure: https://doi.org/10.1016/j.jmb.2010.10.030 and https://doi.org/10.1006/jmbi.1997.1442
But from what I can tell, differences in the sequences there should only have a limited influence on the running of the antibody modeling. Basically, you might pick slightly different templates, but it shouldn't affect the core of the protocol running much. The key thing that will mess the protocol up is things that will affect determination of the CDR loop regions. The Dunbrack and Chothia papers go more into things, but from the perspective of the very N-terminus of the heavy chain, anything before the disulfide cysteine shouldn't influence that.
If you can get away with it, I'd recommend running the protocol with the sequence you actually want to model, rather than modeling it with some other sequence and mutating it later. But it's probably not all that big an issue.