SASREF combined with BUNCH or CORAL

Interactive modelling (MASSHA, SASpy) and global minimization programs (SASREF, BUNCH, CORAL, GLOBSYMM)
Message
Author
james2009
Active member
Posts: 30
Joined: 2009.02.03 16:56
Location: University of Sydney

SASREF combined with BUNCH or CORAL

#1 Post by james2009 » 2011.08.08 16:26

For the next software development of ATSAS, could SASREF can be combined with CORAL or BUNCH?

Where there is an incomplete atomic structure, it is not possible to use this information with SASREF. For small complexes it be less significant, but for large multi-subunits complexes they lack of density is significant.

BUNCH or CORAL can be used separately to model in missing density, but I don't believe it is then possible to use this model in SASREF? Indeed structural changes can occur for the protein in complex

I think there is a need within the community for such a new program (as well as individual symmetries being specified for each phase in the ab initio program MONSA http://www.saxier.org/forum/viewtopic.php?t=705)

Many thanks,

James

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

#2 Post by Alex » 2011.08.08 19:27

Hi James
BUNCH or CORAL can be used separately to model in missing density, but I don't believe it is then possible to use this model in SASREF? Indeed structural changes can occur for the protein in complex
You can 1. model protein with missing parts alone (BUNCH) 2. model protein with missing parts in complex (CORAL). All SAXS modeling is based on the assumption of rigid bodies. So, it is possible to use BUNCH model in SASREF, given it is not changing conformation upon complexation. With CORAL you can assume that conformation of the linker is changing upon complex formation, not the RB.

Well, the whole docking field is trying to model with some success the conformational changes upon complex formation by using either only theoretical approaches or additional constraints (e.g. some experimental data).

At the moment we are trying to model overall changes which occur upon complex formation using NMA and SAXS. Although, it is probably the easiest possible scenario - modeling is not that straightforward, but the first results are encouraging.
It would also be good if SASREF can be combined with CORAL or BUNCH where there is an incomplete structure, it is not possible with SASREF to use an incomplete structure.
Can you explain me why you would not simply use CORAL in this case?

Cheers
Alex

james2009
Active member
Posts: 30
Joined: 2009.02.03 16:56
Location: University of Sydney

#3 Post by james2009 » 2011.08.08 20:11

Hi Alex,

If we have both SANS data which is either protonated/perdeuterated and in different H2O/D2O ratios and SAXS data, I don't think it's possible to use all this different kind of SAS data in CORAL or BUNCH?

So it would be good if CORAL and BUNCH can take into account different contrasts or does it matter where the data is from SAXS or SANS?

Cheers,

James

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

#4 Post by Alex » 2011.08.08 20:17

I don't think it's possible to use all this different kind of SAS data in CORAL or BUNCH?
No, not that I know of.
So it would be good if CORAL and BUNCH can take into account different contrasts or does it matter where the data is from SAXS or SANS?
Yes, it does matter. One has to treat SANS data differently, like you would need to do smearing to take instrument distortion into account, form factors are different etc..

However, I believe it will be implemented in the future version of CORAL.

HTH
Alex

james2009
Active member
Posts: 30
Joined: 2009.02.03 16:56
Location: University of Sydney

#5 Post by james2009 » 2011.08.08 20:31

Ok this would be great.

At the moment I have a protein component with missing density at the C terminus. The protein has six fold symmetry and binds nucleic acid, but the overall complex does not have symmetry, but the protein component does. We have SAXS and SANS data and I am trying to work out the best way to create the ab initio and atomic model. Without there being a P6 symmetry option with MONSA the protein phase does not look like it should, but looks better by CORAL!

What I would really like to do is to let CORAL or BUNCH produce the full-length model for the protein using the atomic model for the protein "core" whilst modelling the missing residues at the C-terminus as dummy residues and then in the absence of an atomic structure for the nucleic acid component, model this as dummy residues.

i.e. there needs to be some kind of combined MONSA/SASREF/CORAL or BUNCH program, which can take into account both SAXS and SANS data +/- deuteration with varying D2O/H2O levels?

Or can you see an alternative more simpler accurate solution? Thanks, James

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

#6 Post by Alex » 2011.08.08 20:45

One of the alternatives, that I can think of, given you have a model for DNA, is to model the missing parts using CORAL with or without contact restraints and SAXS data.

Once a final model(s) is ready, one can cross-validate it using SANS data and CRYSON.

Can you perhaps also send to my email address the SAXS data of the complex for a quick look?

HTH
Alex

Yann
Active member
Posts: 44
Joined: 2010.11.03 11:31
Location: Vrije Universiteit Brussel

Re:

#7 Post by Yann » 2013.09.17 09:02

Hi guys,

I was wondering if there is already any software available for NMA-assisted refinement of structures against SAXS data.
At the moment we are trying to model overall changes which occur upon complex formation using NMA and SAXS. Although, it is probably the easiest possible scenario - modeling is not that straightforward, but the first results are encouraging.
I also found something on the web on NMADREFS... Is this coming out soon, or still under (heavy) construction?

Thanks,

Yann

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: SASREF combined with BUNCH or CORAL

#8 Post by Alex » 2013.09.17 10:56

Hi Yann
I was wondering if there is already any software available for NMA-assisted refinement of structures against SAXS data
I have not seen anything released officially yet.
I also found something on the web on NMADREFS... Is this coming out soon, or still under (heavy) construction?
NMADREFS was meant to be for binary complexes only. I believe if you explain your problem in detail, there may be
people on this forum, including myself, able to help you.

Alex

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: SASREF combined with BUNCH or CORAL

#9 Post by Alex » 2013.09.17 11:23

I think very general explanation can suffice: like - my protein A has a MW of .. and consists of .. domains. I have structure(s) for ..
and I collected SAXS for ..
i tried to model with .. and .. my fit is .. so I want to try NMA-assisted refinement (post your current fit).

A.

Yann
Active member
Posts: 44
Joined: 2010.11.03 11:31
Location: Vrije Universiteit Brussel

Re: SASREF combined with BUNCH or CORAL

#10 Post by Yann » 2013.09.17 11:29

Hi Alex,

thanks for your quick reply.

Yes, I'll explain my problem. I have a crystal structure of a complex comprising 16 subunits: it is hexadecamer built up of 8 heterodimers. All of them have affinity tags, which are not visible in the structure, but are there. One partner has an N-terminal FLAG-tag and the other one a C-terminal His-tag. So in total: 16 tags flopping around.

When I compute the scattering curve with crysol on the structure without the tags, I have a chi of 11.672 (see Fit1).

I then plugged my structure into the AllosMod-Foxs server and got a significant improvement: chi of 2.569 (see Fit2). This is the result for just one structure with the tags.

However, there is room for improvement. I would like to try to fit with multiple 'tag'-conformations. Also, I suspect some movement within the complex between the different subunits in solution. I already plugged the crystal structure (no tags) in NOMAD-REF and got some interesting results.

I already tries MES on the FOXS server, and OLIGOMER, but they both give the weight to one conformer so I am stuck at 2.569.

Any pointers on how I could improve my fit with this info?

Thanks,

Yann
Attachments
Fit1.jpg
Fit1.jpg (27.7 KiB) Viewed 8047 times
Fit2.jpg
Fit2.jpg (26.16 KiB) Viewed 8047 times

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: SASREF combined with BUNCH or CORAL

#11 Post by Alex » 2013.09.17 11:56

Yes, I'll explain my problem. I have a crystal structure of a complex comprising 16 subunits: it is hexadecamer built up of 8 heterodimers. All of them have affinity tags, which are not visible in the structure, but are there. One partner has an N-terminal FLAG-tag and the other one a C-terminal His-tag. So in total: 16 tags flopping around.
That is a big guy!
Bear in mind, that it would make no sense to (NMA)-refine a crystal structure without tags against data corresponding to a construct with a tag. Although perhaps the protein can be much larger as compared to the tag, the amino acids in the tag has aromatic ring which contributes significantly to scattering. All 16 tags have to be present should you want to refine using NMA.

Do the tags point at neighboring subunit(s) or located very close to it? Should you want to sample many possible conformations of the tags, this should be taken into account to avoid steric clashes.
I then plugged my structure into the AllosMod-Foxs server and got a significant improvement: chi of 2.569 (see Fit2). This is the result for just one structure with the tags.
If i understand it right - you modeled tags using AllosMod, which uses Modeller to reconstruct missing fragment(s), right?
Did you try CORAL to do it too? What was a CRYSOL fit if you take this final structure and fit to exp data?
Did you try new FOXS with explicit water modeling? If so, is that the fit you posted?

What SAXS data are you using (concentration, extrapolated?, where was it recorded)?
From what I see on fit 2, the misfit indeed maybe due to internal motions of the protein and could be refined with
NMA approach.


HTH
Alex

Yann
Active member
Posts: 44
Joined: 2010.11.03 11:31
Location: Vrije Universiteit Brussel

Re: SASREF combined with BUNCH or CORAL

#12 Post by Yann » 2013.09.17 12:12

Hi Alex,

big indeed!

AllosMod indeed uses MODELLER and I used FOXS with the default settings (uses hydration layer to improve fit). I tried to use CORAL, but it always stopped with the error message 'STOP cannot build C-terminal'. Don't know what went wrong there...

The data set was recorded at the SOLEIL synchrotron using the HPLC set-up. I injected about 80ul at 10 mg/ml, and shot the peak. All curves were individually analyzed, and the merged to obtain this final curve.

The tags should point outwards, although the tag of partner A is indeed very close to partner B and vice versa, so clashes should definitely be avoided.

I agree with your comment on the NMA on tag-less protein. I tried to perform NMA on the AllosMod structure, but then I only got big movements in one of the tags.

Looking forward to hearing your thoughts.

Thanks,

Yann

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: SASREF combined with BUNCH or CORAL

#13 Post by Alex » 2013.09.17 12:24

AllosMod indeed uses MODELLER and I used FOXS with the default settings (uses hydration layer to improve fit). I tried to use CORAL, but it always stopped with the error message 'STOP cannot build C-terminal'. Don't know what went wrong there...
It is peculiar that CORAL failed. You should post the required files here so people can reproduce your problem with CORAL and help you.
The data set was recorded at the SOLEIL synchrotron using the HPLC set-up. I injected about 80ul at 10 mg/ml, and shot the peak. All curves were individually analyzed, and the merged to obtain this final curve.
What is the MW of the complex? What was the concentration of the complex at the peak you shot?

Alex
Active member
Posts: 692
Joined: 2007.08.09 21:10
Location: Planet Earth

Re: SASREF combined with BUNCH or CORAL

#14 Post by Alex » 2013.09.17 12:27

if you also post your model with tags, I can give a try to my refinement scripts/or suggest a better strategy.
Alex

Yann
Active member
Posts: 44
Joined: 2010.11.03 11:31
Location: Vrije Universiteit Brussel

Re: SASREF combined with BUNCH or CORAL

#15 Post by Yann » 2013.09.17 12:30

Hi Alex,

just post my pdb here or send you a mail with my pdb file?

Thanks a bunch!

Yann

PS: Guess you'll also need my experimental data

Post Reply