Hello,
I am trying to model a multichain protein using 2 templates - the .pdb file of an oligomer (with the same number of chains as the target) and the .pdb file of the AlphaFold structure of one of the monomers making up the target protein. (When I model using only the oligomer as target, the model of the target looks quite "off".)
First scenario: * target protein: 5 of the same chain * template protein 1 ("base template"): a homopentamer * template protein 2 ("sub-template" - i.e. the target protein is first modelled based on the first template then modelled based on the 2nd template): the AF predicted structure of one chain of the target protein (the .pdb file has only one chain).
Second scenario: * target protein: a heteropentamer - 2 of one chain + 3 of another chain * template protein 1: as in the previous scenario * template protein 2 (+3): AF predicted structures of the two chains making up the target heteropentamer. There are 2 AF folds, one for each monomer, each one with a single chain.
How would I format the alignment file?
I look forward to your guidance.
Kind regards, Amanda
Apologies - I just found out that someone else asked basically the same question months prior: https://salilab.org/mm/hyperkitty/list/modeller_usage@salilab.org/thread/353...
I presume that in the second scenario, I'd have three templates, a la: temp1 AAAA/AAAA/BBBB/BBBB/BBBB* temp2 aaaa/----/----/----/----* temp3 ----/-----/bbbb/----/----* model aaaa/aaaa/bbbb/bbbb/bbbb*
What if I am modelling with ligands? Would the following be right? temp1 AAAA./AAAA/BBBB./BBBB/BBBB* temp2 aaaa-/----/----/----/----* temp3 ----/-----/bbbb-/----/----* model aaaa./aaaa/bbbb./bbbb/bbbb*
(The .pdb files for temp2 and temp3 have no ligands. Should the ligand be indicated? If so, with a "-" or "."?)
On 3/19/24 8:38 AM, hmad3--- via modeller_usage wrote: > What if I am modelling with ligands? Would the following be right? > temp1 AAAA./AAAA/BBBB./BBBB/BBBB* > temp2 aaaa-/----/----/----/----* > temp3 ----/-----/bbbb-/----/----* > model aaaa./aaaa/bbbb./bbbb/bbbb*
Yes. Normally I would align the chain breaks with each other to make it more clear to a human reader, but Modeller doesn't distinguish a chain break from a gap in a template.
> (The .pdb files for temp2 and temp3 have no ligands. Should the > ligand be indicated? If so, with a "-" or "."?)
The sequence for a template must match what's in the PDB file. So if there is no ligand, there should be no ".". Normally you would put a gap here so that the target does not align with anything else in that template at that position. (Generally it only makes sense for a target ligand to align with a single template ligand.)
Ben Webb, Modeller Caretaker
Thanks a lot for your reply.
Maybe I'm being a bit over-cautious, but if I'd like all of my subunits to be aligned according to the AF model(s), would there need to be n (where n = number of subunits) templates in the .ali file alongside the template corresponding to the pentameric .pdb file? Something like:
temp1 AAAA./AAAA/BBBB./BBBB* temp2 aaaa-/----/-----/----* (note the 5 dashes in the third "chain") temp3 -----/aaaa/-----/----* (5 dashes in the 1st and 3rd "chain") temp4 -----/----/bbbb./----* (5 dashes in the first "chain") temp5 -----/----/-----/bbbb* model aaaa/aaaa/bbbb/bbbb/bbbb*
On 3/19/24 11:12 AM, hmad3--- via modeller_usage wrote: > Maybe I'm being a bit over-cautious, but if I'd like all of my > subunits to be aligned according to the AF model(s), would there need to > be n (where n = number of subunits) templates in the .ali file alongside > the template corresponding to the pentameric .pdb file?
Yes, absolutely. Modeller will use exactly the alignment you provide it, so if you want your monomer to be used to model every chain, you should provide multiple copies; Modeller will not assume that otherwise. The alternative would be to add a symmetry restraint to force every chain to look alike: https://salilab.org/modeller/10.5/manual/node29.html
Ben Webb, Modeller Caretaker
Dear Ben, Thank you very much. One last question (I'm pretty new to Modeller) about the .ali file and Python code:
* I presume that for the .ali file, the header for the AF sequences would be >P1;temp2 structureX:temp2::A::A::::
and what would the part of the Python code that looks like: a = automodel(env, alnfile = 'alignment.ali' , # alignment filename knowns = 'temp1', # codes of the templates sequence = 'target') # code of the target need to look like? Would there need to be a list/array for "knowns"?
With kind regards, Amanda
Apologies, I've found the answer via a quick Googling (d'oh!).
-
hmad3＠cam.ac.uk -
Modeller Caretaker