SAXS

Small Angle X-ray Scattering Initiative for Europe :: Forum
It is currently 2017.11.18 04:30

All times are UTC + 2 hours




Post new topic Reply to topic  [ 13 posts ] 
Author Message
PostPosted: 2017.09.07 08:20 
Offline
Member

Joined: 2015.11.23 14:19
Posts: 3
We would like to use EOM to support some rigid body modelling we did with CORAL. We would like to use EOM to make a pool of PDBs for a molecule with two well-defined structures connected by a flexible linker (see attached cartoon of the model). Internally both of the two structures have P2 symmetry which is very well defined, and which we want to keep fixed. The overall structure does not have perfect P2 symmetry. When using EOM/Ranch we are unable to make the software generate this kind of structure. Either the program will not run, or it pulls apart the P2 symmetry of one of the well-defined structures.

Is it possible to make this kind of model using EOM?


Attachments:
File comment: system we would like to model
model.png
model.png [ 7.18 KiB | Viewed 408 times ]
Top
 Profile  
 
PostPosted: 2017.09.12 14:38 
Offline
Administrator
User avatar

Joined: 2007.08.03 18:55
Posts: 608
Location: EMBL Hamburg, Germany
Okay, let's try the following. Take dimer models of PDB1 and PDB2, put them to origin (e.g. with ALPRAXIN). Split PDB1 dimer into two files each containing a monomer: 'PDB1a.pdb' and 'PDB1b.pdb'.

Let's assume the sequence of PDB1 is:
WAAAAAAAX
The sequence of PDB2 is:
YBBBBBZ
The sequence of the linker is
LLL

Let's assume the monomeric full-length sequence is
WAAAAAAAXLLLYBBBBBZ

Open 'PDB1b.pdb' in a text editor, cut the last residue (X) and paste it to the end of the file.
Open 'PDB2-dimer.pdb' in a text editor, cut the first residue (Y) of the second chain and paste it to the end of the file.

Prepare the following 'sequence.seq' file:
WAAAAAAAX
LLL
YBBBBBZ
BBBBBZY
LLL
XWAAAAAAA

Run EOM online with the following configuration:

Overall symmetry: P1
Sequence: 'sequence.seq'
Number of domains: 3
Domain 1: 'PDB1a.pdb' [monomer] [fixed]
Domain 2: 'PDB2-dimer.pdb' [monomer] [free]
Domain 3: 'PDB1b.pdb' [monomer] [fixed]

If you want to publish one of the resulting models you may want to move the residues X and Y back to where they belong in the original sequence.


Top
 Profile  
 
PostPosted: 2017.09.14 09:03 
Offline
Member

Joined: 2015.11.23 14:19
Posts: 3
Thank you for your time and energy! It seems to be running as it is supposed to.


Top
 Profile  
 
PostPosted: 2017.09.25 11:44 
Offline
Active member

Joined: 2015.11.03 14:38
Posts: 24
Hi,

I thought I would be ok to continue your topic since I'm encountering a sort-of similar issue. I have a protein composed of 3 domains. We know it forms dimers in solution and we suppose it's through domain A (it is the case in proteins of the same family which we have structure from). We have SEC-SAXS data on the dimer peak.
I used this approach: get the model of the dimer domain A (from homologous protein) + model of domain B + model of domain C as separate pdb's. Use them with domain A as fixed multimer + overall P2 symmetry.

The program does run, but the Chi2 is about 43-48 (very high) and the fit (attached) looks super bad and weird. Models don't look bad but well...

A guess on why the fit is so bad and how to fix/process this?
Many thanks!


Attachments:
profiles_001_1.fit [48.76 KiB]
Downloaded 4 times
Top
 Profile  
 
 Post subject: Not that flexible?
PostPosted: 2017.09.26 15:24 
Offline
Administrator
User avatar

Joined: 2007.08.03 18:55
Posts: 608
Location: EMBL Hamburg, Germany
Nyshae wrote:
We know it forms dimers in solution and we suppose it's through domain A (it is the case in proteins of the same family which we have structure from). We have SEC-SAXS data on the dimer peak.
And the expected MW of your dimer is around 200 kDa, correct?

Nyshae wrote:
... the Chi2 is about 43-48 (very high) and the fit (attached) looks super bad and weird.
This means it is not possible to fit your data with the provided domains and the given degree of flexibility. Maybe your protein is not that flexible?
dara.embl-hamburg.de suggests that your protein looks like the biological assembly 3 of PDB ID: 3THO (169 kDa dimer). Have you tried modelling your protein with CORAL?


Top
 Profile  
 
PostPosted: 2017.09.28 09:25 
Offline
Active member

Joined: 2015.11.03 14:38
Posts: 24
Thanks for the reply!

Well actually each monomer is around 52 kDa + 4 glycosylations and on MALS it runs as a 106-120 kDa species (I don't have the precise weight right now, it's another persons' part). So, smaller than what you say... How did you get the 200?

We think it might be somewhat flexible because between domain B and C there's a 39 AA linker with no predicted structure.

I did try to run CORAL, with different options:
-online version with each domain as monomer model = Chi2 around 4, fit visually ok but ABC dimer far apart
-shell version with each domain as monomer model + configuration file for linker length + contact file for the dimerization of domain A (I played a bit with different length and AA) = same result
-shell version with the dimer of domain A + domain B + domain C as seeds + configuration file indicating the linkers in between them = Chi2 around 1, visually good fit but the problem is that CORAL is attaching the domains C a bit randomly (not from the right place, one is from domain A and the other from domain B but wrong end). I think by default CORAL is attaching domains C to the extremities of the dimer pdb, which of course are Cter of A on one side and Nter of A' on the other and I don't really see how to change that...

I am also using individual pdb models with glycans added for CORAL (for EOM it wasn't working so I removed them). Could it be a problem?


Top
 Profile  
 
 Post subject: CORAL online: dimer
PostPosted: 2017.09.28 13:50 
Offline
Administrator
User avatar

Joined: 2007.08.03 18:55
Posts: 608
Location: EMBL Hamburg, Germany
Nyshae wrote:
How did you get the 200?
Porod volume divided by 1.6 (it might be overestimated for elongated/flexible proteins). DARA neighbours are around 160 kDa.
Nyshae wrote:
We have SEC-SAXS data on the dimer peak.
I assume the peak was well separated. I hope there was another peak or two.

Nyshae wrote:
I did try to run CORAL, with different options:
-online version with each domain as monomer model = Chi2 around 4, fit visually ok but ABC dimer far apart...
Take the dimer model of domain A, put it to origin (e.g. with ALPRAXIN) and make sure the symmetry axis is aligned with z. Then remove one monomer from the pdb file. Run CORAL online with the following configuration:
    Overall symmetry: P2
    Number of domains: 3
    Domain 1: A-monomer.pdb [fixed] [linker] [length]
    Domain 2: B-monomer.pdb [free] [linker] [39]
    Domain 3: C-monomer.pdb [free] [C-terminal]
Once it starts make sure the dimer of domain A was reconstructed properly.

Nyshae wrote:
I am also using individual pdb models with glycans added for CORAL (for EOM it wasn't working so I removed them). Could it be a problem?
No, I don't think this is a problem for CORAL.


Top
 Profile  
 
PostPosted: 2017.09.28 16:15 
Offline
Active member

Joined: 2015.11.03 14:38
Posts: 24
So after verification, the dimer is 106 kDa. There is indeed another peak, suspected tetramer (212 kDa). They unfortunately do overlap a little bit... I tried to select and average the frames that had a stable Rg (rather the end of the dimer peak), I hope it's enough...

Thanks a lot for your suggestion, I'll try that right away. It might seems like a newbie question but how do you "make sure that the symmetry axis is aligned with Z" on PyMol? I tried the "orient" function, the dimer does move but I doubt it's properly aligned... (no display of the axis)


Top
 Profile  
 
 Post subject: ALPRAXIN + PyMOL
PostPosted: 2017.09.29 15:01 
Offline
Administrator
User avatar

Joined: 2007.08.03 18:55
Posts: 608
Location: EMBL Hamburg, Germany
Nyshae wrote:
...how do you "make sure that the symmetry axis is aligned with Z" on PyMol?
Make sure both monomers are (almost) identical. Run ALPRAXIN on the dimer (either from the SASpy plugin for PyMOL or from the command line). Check the rotated model in PyMOL - it is not unlikely that it will be properly oriented (you are looking at it along the z-axis). If you don't see the two dimers in the same plane - rotate by 90° using the rotate command.


Top
 Profile  
 
PostPosted: 2017.10.02 10:06 
Offline
Active member

Joined: 2015.11.03 14:38
Posts: 24
Hi AL,

So I managed to do what you suggested and it worked very nicely! My Chi2 is not super good (around 9) and the fit is not either but it seems much better than previously!

I suspect the "not super goodness" might be due to the fact that the frames I used are not 100% monodisperse (I used CORAL on a shorter version of the same protein, with unique peak in SEC-SAXS, and Chi is around 1-2)... Any ideas on how I could maybe "sub-select" a better frame range?

Thanks a lot!


Top
 Profile  
 
PostPosted: 2017.10.02 15:27 
Offline
Administrator
User avatar

Joined: 2007.08.03 18:55
Posts: 608
Location: EMBL Hamburg, Germany
Nyshae wrote:
I suspect the "not super goodness" might be due to the fact that the frames I used are not 100% monodisperse (...) Any ideas on how I could maybe "sub-select" a better frame range?
You can open your frames in CHROMIXS and try to manually select only one side of the peak. Of course it would be better to collect SEC-SAXS data using a better column.
You may try SASREF MX (choose "SAXS only, equilibrium mixture" in SASREF online) or EOM with two pools (dimers and tetramers).


Top
 Profile  
 
PostPosted: 2017.10.04 11:15 
Offline
Active member

Joined: 2015.11.03 14:38
Posts: 24
Thanks! I'll dig in a bit more then.


Top
 Profile  
 
PostPosted: 2017.10.20 11:06 
Offline
Member

Joined: 2009.01.12 10:55
Posts: 13
Location: IBCP,france
Dear All
I got a similar problem of two models with internal P2. But this time with a relatively long IDR at the Nter.
from the picture you can see the organisation of my protein:
Attachment:
eom-dim.png
eom-dim.png [ 8.77 KiB | Viewed 205 times ]

1- Nter linker (50aa),
2- first dimer interface with the pdb.
3- then another linker (110 aa)
4- finally the last dimer

By following the nice step to step protocol from AL. I succeed to obtains models but without the Nter.
I didn't find the trick to make ranch generate the whole protein.
If you have any clue how to generate such model it will be a relief for me .
Cheers
JM


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 13 posts ] 

All times are UTC + 2 hours


Who is online

Users browsing this forum: No registered users and 2 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Powered by phpBB® Forum Software © phpBB Group