New web server for SAXS/WAXS calculations

Small angle scattering software (except ATSAS)
Post Reply
Message
Author
jochenhub
Member
Posts: 4
Joined: 2014.06.30 15:30

New web server for SAXS/WAXS calculations

#1 Post by jochenhub » 2014.06.30 15:37

Dear SAXS community,

it is our pleasure to announce a new web server for the calculation and fitting of SAXS/WAXS curves of biomolecules in solution.

WAXSiS ("WAXS in Solvent") computes SAXS/WAXS curves based on explicit-solvent all-atom molecular dynamics simulations.

The main differences to established web services are:

* No fitting parameters associated with the solvation shell or excluded solvent are required.
* Solvation shell and excluded solvent are described at atomic detail, allowing accurate scattering predictions also at wide angles.
* Thermal fluctuations of solvent and biomolecule are naturally included.

Please find WAXSiS at:

http://waxsis.uni-goettingen.de/

WAXSiS is up and running, and it has passed a few hundred test calculations. We would be pleased to receive
feed-back, comments, and functionality requests from the SAXS community.

With best wishes,
The WAXSiS team at the University of Göttingen

rambor
Active member
Posts: 27
Joined: 2008.02.29 03:29
Location: USA

Re: New web server for SAXS/WAXS calculations

#2 Post by rambor » 2014.07.01 12:20

I tried and it failed after two hours (using single chain from P4-P6 RNA domain crystal structure). 2 hours is quite long to notify someone it didn't work

So far: CRYSOL 1, FOXS 1, WAXSiS 0

jochenhub
Member
Posts: 4
Joined: 2014.06.30 15:30

Re: New web server for SAXS/WAXS calculations

#3 Post by jochenhub » 2014.07.01 15:49

rambor wrote:I tried and it failed after two hours (using single chain from P4-P6 RNA domain crystal structure). 2 hours is quite long to notify someone it didn't work

So far: CRYSOL 1, FOXS 1, WAXSiS 0
Dear rambor,

thank you for reporting this. We fixed the error, and your job should work now. - So there is hope for an equalizer with Crysol and Foxs in extra time. :)

But please note that WAXSiS not designed for speed, but instead to model the solvation shell accurately, without the requirement to fit the solvent contributions.

Best,
Jochen

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

Re: New web server for SAXS/WAXS calculations

#4 Post by Alex » 2014.07.01 17:33

as the full text of the article is not available yet, can you comment if WAXSIS outperforms CRYSOL, FOXS and AXES (in terms of speed and goodness of fit)?
thanks, Alex

rambor
Active member
Posts: 27
Joined: 2008.02.29 03:29
Location: USA

Re: New web server for SAXS/WAXS calculations

#5 Post by rambor » 2014.07.01 22:46

Thanks Jochen,


The results were very nice. Have you seen Peter Moore's recent paper in Biophysical journal regarding corrections to WAXS data?

CRYSOL 1, FOXS 1, WAXSiS 1

jochenhub
Member
Posts: 4
Joined: 2014.06.30 15:30

Re: New web server for SAXS/WAXS calculations

#6 Post by jochenhub » 2014.07.03 15:02

Alex wrote:as the full text of the article is not available yet, can you comment if WAXSIS outperforms CRYSOL, FOXS and AXES (in terms of speed and goodness of fit)?
thanks, Alex
Dear Alex,

in short, the philosophy behind WAXSiS is

(a) to model the solvation shell and excluded solvent accurately using explicit-water MD simulations (with the aim to avoid fitting of solvent contributions),

(b) to include atomic fluctuations, which are relevant at wide angles.

Since we run an MD simulation for that, the calculations take much longer compared to Foxs or Crysol. The MD simulation takes roughly 30 Minutes on six (quite slow) AMD Opteron cores, while the actual WAXS calculation takes a few more minutes. That's why WAXSiS requires slightly more patience.

Regarding the goodness of fit, we cannot simply compare chi2 values because we use (apart form the overall intensity scale) only one fitting parameter c, that aims to absorb experimental uncertainties from the buffer subtraction:

I(fitted) = f * I(experiment) + c

Consequently, the chi^2 may be similar or larger compared to other methods that use more fitting parameters, but the fitting procedure should be more sensitive with respect to smaller conformational transitions of the solute. We are currently running more test to actually quantify that.

Until the paper is out, there is some basic background also in the About section of the server:

http://waxsis.uni-goettingen.de/about/

Best wishes,
Jochen

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

Re: New web server for SAXS/WAXS calculations

#7 Post by Alex » 2014.07.03 15:10

Thanks for explanation! I give it a try soon.
Alex

jochenhub
Member
Posts: 4
Joined: 2014.06.30 15:30

Re: New web server for SAXS/WAXS calculations

#8 Post by jochenhub » 2014.07.03 15:12

rambor wrote:Thanks Jochen,

The results were very nice. Have you seen Peter Moore's recent paper in Biophysical journal regarding corrections to WAXS data?

CRYSOL 1, FOXS 1, WAXSiS 1
Hi ramor,

thanks for pointing to this paper, which I missed so far (came out just after we submitted the major revision or our paper). The paper shows very clearly that atomic fluctuations (and the correlations between those fluctuations) are relevant in the WAXS regime.

Actually that's good news, since WAXSiS incorporates such fluctuations (and correlations) - to the extent present in MD simulations with positions restraints on the backbone. Notably, as described in the paper, we compared WAXS curves from simulations with different degrees of flexibility, and position-restrained simulations seemed good enough.

Best,
Jochen

Post Reply