EOM on a homo-oligiomer

Linear (OLIGOMER), and non-linear (MIXTURE) analysis, singular value decomposition (SVDPLOT), addition of missing fragments (BUNCH, CORAL), analysis of flexible systems (EOM/RANCH & GAJOE), flexible refinement of high-resolution models (SREFLEX)
Post Reply
Message
Author
rileydm
Member
Posts: 5
Joined: 2018.02.20 07:34

EOM on a homo-oligiomer

#1 Post by rileydm » 2018.02.20 08:26

Hi all,

We have the crystal structure of a homo-heptamer, and we have collected SAXS data on this protein. There is a large (~50 residue) missing internal loop in the structure, protruding from the core of the heptameter. I have modelled this loop using CORAL, but we're interested in the flexibility and ensemble of this loop, so I'm hoping to use EOM.

Is there a way to use the P7 symmetry to re-construct the heptameter from the monomer, similar to CORAL? I define each of the structured parts as a 'domain' in EOM, fix them in place, similar to CORAL, but EOM won't run without a contact conditions file, and when this is provided, won't reconstruct the interface properly.

Any pointers?

Edit: What I'm trying to do is similar to the example given here (https://www.embl-hamburg.de/biosaxs/cou ... ertens.pdf), or figure 2 in this paper (http://journals.iucr.org/m/issues/2015/ ... fc5007.pdf) - just with two rigid bodies per subunit instead of three.

User avatar
AL
Administrator
Posts: 648
Joined: 2007.08.03 18:55
Location: EMBL Hamburg, Germany
Contact:

Re: EOM on a homo-oligiomer

#2 Post by AL » 2018.02.21 14:42

Align your heptamer with ALPRAXIN.

Open the aligned pdb file in a text editor, leave one monomer (any), remove the other six. Further split this monomer into two domains which should be connected by the internal flexible loop.

Run EOM online:
  • Overall symmetry: P7
    Sequence: [monomeric sequence]
    Number of domains: 2
    Domain 1: [monomer][fixed]
    Domain 2: [monomer][fixed]

rileydm
Member
Posts: 5
Joined: 2018.02.20 07:34

Re: EOM on a homo-oligiomer

#3 Post by rileydm » 2018.02.21 15:28

Thanks Al,

Tried that, but gives an error 'Due to symmetry P7, at least on of the domains must be an oligomer,' immediately after submitting it.

User avatar
AL
Administrator
Posts: 648
Joined: 2007.08.03 18:55
Location: EMBL Hamburg, Germany
Contact:

Re: EOM on a homo-oligiomer

#4 Post by AL » 2018.02.22 16:28

Then try providing the domain with the oligomerization interface as a heptamer and the other domain as a monomer:
  • Overall symmetry: P7
    Sequence: [monomeric sequence]
    Number of domains: 2
    Domain 1: [multimer][fixed]
    Domain 2: [monomer][fixed]

rileydm
Member
Posts: 5
Joined: 2018.02.20 07:34

Re: EOM on a homo-oligiomer

#5 Post by rileydm » 2018.02.23 14:10

Thanks Al,

It's running now, but one problem has been replaced by another... Gives the error message below (EOM online run untitled_3145):

Code: Select all

 --------------------- ATTENTION! ---------------------
 - The domains specified as fixed may be too far away -
 - to be connected by the linker extracted from the   -
 - sequence. EOM 2.1 may not be able to make the pool.-
 ------------------------------------------------------
 Alignment Problems: check the sequence and the domains
find: `./GA001/curve_1/Rg_*.txt': No such file or directory
find: `./GA001/curve_1/Size_*.txt': No such file or directory
find: `./GA001/curve_1/*.fit': No such file or directory
When run locally, starts okay but there's no error and RANCH doesn't start generating models.

I've used pdb2seq to check that the correct sequence is in the sequence file, and tried the method suggested in a previous thread to ensure that the correct sequence is provided. In case it's an issue of the linker being too short (which seems unlikely) - I've increased the number of alanines in the linker, to try and overcome it, to still no avail. I've also tried uploading both domains as a multimer (and still doesn't work, as it run untitled_3146). Any pointers? Are there any requirements for the sequence file beyond plaintext, single letter residue code (as indicated in the manual?)

Thanks again for your help - apologies if I'm being dense!

User avatar
AL
Administrator
Posts: 648
Joined: 2007.08.03 18:55
Location: EMBL Hamburg, Germany
Contact:

Re: The domains specified as fixed may be too far away

#6 Post by AL » 2018.03.05 15:47

rileydm wrote:Are there any requirements for the sequence file beyond plaintext, single letter residue code (as indicated in the manual?)
Hm, not really.

Does EOM run if you change to
  • Domain 1: [multimer][fixed]
    Domain 2: [monomer][free]?

rileydm
Member
Posts: 5
Joined: 2018.02.20 07:34

Re: The domains specified as fixed may be too far away

#7 Post by rileydm » 2018.04.12 10:58

AL wrote:
rileydm wrote:Are there any requirements for the sequence file beyond plaintext, single letter residue code (as indicated in the manual?)
Hm, not really.

Does EOM run if you change to
  • Domain 1: [multimer][fixed]
    Domain 2: [monomer][free]?
Thanks Al. Sorry about the delay, been trying again to get this working recently.

Running:

Domain 1: [multimer][fixed]
Doman 2: [monomer][free]

(and vice-versa) works - RANCH starts to generate the pool, although a lot of the models are unsurprisingly nonsensical, so it doesn't seem to be a problem with the sequence. To generate the multimeric PDBs, I have overlaid chain A in pymol to the symmetry-related chains in the PDB, to ensure that all chains in the multimeric PDB that I use as an input for EOM are identical (there are subtle chain-to-chain differences in the crystal structure).

Running:

Domain 1: [multimer][fixed]
Domain 2: [monomer][fixed]

(and vice versa) doesn't work. RANCH starts, but gives an error that the termini may be too far apart to be joined by the linker extracted from the sequence. The 'termini' are close in space (within 10 Å) joined by a long linker (~50 residues long) which extends out into the solvent - the flexibility of this is what we are interested in modelling.

Finally, running:

Domain 1: [monomer][fixed]
Domain 2: [monomer][fixed]

on EOM Online gives an error that a multimeric interface must be defined. Running it locally, it won't run unless a contacts file is given (which I understand is working as designed).

User avatar
AL
Administrator
Posts: 648
Joined: 2007.08.03 18:55
Location: EMBL Hamburg, Germany
Contact:

Re: The domains specified as fixed may be too far away

#8 Post by AL » 2018.04.25 18:21

Tried it myself on my old heptamer project - indeed if both domains are fixed ranch says "The domains specified as fixed may be too far away" and hangs. I can generate models with fixed domains and flexible termini but flexible internal linkers seem to be impossible in EOM. I've opened a problem report on this issue.
rileydm wrote:Running it locally, it won't run unless a contacts file is given (which I understand is working as designed).
Yes, if a symmetry is defined but all domains are monomers - EOM requires a contacts file. Unfortunately it is not possible to use this file to fix the relative orientation of the two domains.

One could write a script that runs BUNCH multiple times (perhaps making just one simulated annealing step for each time) and use the resulting models as a pool for EOM/GAJOE.

rileydm
Member
Posts: 5
Joined: 2018.02.20 07:34

Re: The domains specified as fixed may be too far away

#9 Post by rileydm » 2018.04.26 01:30

AL wrote:Tried it myself on my old heptamer project - indeed if both domains are fixed ranch says "The domains specified as fixed may be too far away" and hangs. I can generate models with fixed domains and flexible termini but flexible internal linkers seem to be impossible in EOM. I've opened a problem report on this issue.
rileydm wrote:Running it locally, it won't run unless a contacts file is given (which I understand is working as designed).
Yes, if a symmetry is defined but all domains are monomers - EOM requires a contacts file. Unfortunately it is not possible to use this file to fix the relative orientation of the two domains.

One could write a script that runs BUNCH multiple times (perhaps making just one simulated annealing step for each time) and use the resulting models as a pool for EOM/GAJOE.
Thanks for looking into this Al, might give that a try.

Let me know if you find a solution for RANCH - I'll be happy to give it a shot!

Post Reply