Page 1 of 1

Help with GNOM/Gasbor

Posted: 2018.08.02 23:00
by oavina24
Untitled.png (105.95 KiB) Viewed 5826 times
Can someone walk me through the steps of getting data into and out of GNOM and into Gasbor. The samples are of HSA 5mg/ml taken in a xeuss 2.0 beamline at 1209 sd, high resolution for 1200 seconds. The files were circle gathered using foxtrot and then normalized and background subtracted using mantid plot. Now I have .dat files and would like to do some ab-initio modeling, the probelm is that some of the samples are at higher temperatures, and as I understand Dammif is not very good with unfolded systems and Gasbor is more appropriate.

Use PRIMUSqt and GASBOR online

Posted: 2018.08.03 11:25
by AL
1. Make sure you have the latest ATSAS version installed.
2. Open your background-subtracted data in PRIMUS (on Windows make sure you use primusqt; it is called "SAS Data Analysis" in the Start menu).
3. In the bottom right panel go to the 'Analysis' tab and click Distance Distribution'. This will start the GNOM GUI.
3.1. If PRIMUS was unable to automatically estimate Rg from the Guinier approximation there will be an additional step where you have to choose a Guinier region manually.
4. Increase the last point of 'Range' such that the whole experimental data are displayed (you can look up the number of data points in the parent window under 'To').
5. If necessary - adjust 'Dmax'; make sure your p(r) is smooth, positive and fits the experimental data.
6. 'Save' your p(r) and the fit to the experimental data in a file with a *.out extension.
7. Go to GASBOR online.
  • 'GNOM file': choose your *.out file;
  • 'Angular units': choose inverse nanometres (1/nm) or inverse angstroms (1/A).
  • 'Symmetry': if the molecular weight estimated from the SAXS data points to a monomer - choose 'P1'.
  • 'Anisometry': if you think your protein is rather elongated than flat - choose 'Prolate', otherwise leave 'Unknown'.
  • 'Number of dummy residues': number of residues in the protein you've measured (including His-tag etc.).
  • 'Mode': 'Reciprocal space'. (If you have evidence your protein is a monomer:dimer mixture in solution - you may try 'Symmetry: P2' and choose 'Mode: oligomer-monomer mixture'.)
8. Once GASBOR is finished - check the output files.
Run GASBOR 5-10 times to make sure that you always get about the same model and all your models fit your input data. To do so go to My Projects and click 're-run with new id' several times.
9. If the fit is good and the model makes sense - deposit it to SASBDB.

Check out the EMBO practical course on solution scattering from biological macromolecules 19 - 26 November 2018 | Hamburg, Germany

Re: Hel p with GNOM/Gasbor

Posted: 2018.08.03 22:52
by oavina24
Thank you this really helps, but i still have questions. In the distance distribution analysis tab I have total quality estimate which I guess is a percentage value. Guinier Rg/I(0) , p(r) Rc/I(0) and porod volume, what are these values and what should they be. For system I'm using a Xeuss 2.0 by xenocs in High resolution. How do I know the range and what system im dealing with, I know I'm using HSA. What does Rmin=0 and Rmax=0 mean and points and Alpha. I know it's a lot of questions but i would greatly appreciate it.

GNOM parameters

Posted: 2018.08.07 17:26
by AL
  • Left plot log(I) vs. s: fit of the Fourier transform of p(r) (blue line) to the experimental SAS data (black dots).
    The fit is computed like this: I(s) = 4Π ∫oDmax p(r)sin(sr)/sr dr
  • Right plot p(r) vs. r: pair distance distribution function.
  • Total quality estimate ranges between 1.0 (best) and 0.0 (worst).
  • Guinier Rg/I(0): Radius of gyration and Forward scattering estimated from the fit in the left plot using Guinier approximation.
  • p(r) Rg/I(0): Radius of gyration and Forward scattering estimated from the pair distance distribution function (right plot).
    If the experimental data are in inverse angstroms (Å-1), Rg is provided in angstroms (Å). I(0) is provided in units of I(s).
  • Porod Volume: Excluded volume of the hydrated particle computed by DATPOROD.
    If the experimental data are in inverse angstroms (Å-1), the volume is provided in cubic angstroms (Å3).
    For globular proteins Porod volume in cubic angstroms divided by 1.6 roughly corresponds to the molecular weight in daltons (Da).
  • System: Arbitrary Monodisperse - monodisperse systems of globular particles; GNOM can treat experimental data from other types of systems.
  • Range: first and last points of the experimental data used by GNOM. By default the first point is the first point of the Guinier interval, the last point is s = 8/Rg.
  • Setup: experimental setup, either Point Collimation (no smearing) or Slit Collimation.
  • Dmax: maximum intra-particle distance.
  • Rmin=0: force r(0) = 0 (typically you want this).
  • Rmax=0: force r(Dmax) = 0 (typically you want this).
  • Points: number of p(r) data points.
  • Alpha: regularization parameter.
    Larger alpha values yield a smoother p(r) but worse fit to the experimental data.
    Smaller alpha values improve the fit but may result in a wobbly p(r).
  • Autognom: run DATGNOM and return to the initial state of this Wizard (if you've changed any values).

Re: Help with GNOM/Gasbor

Posted: 2020.02.20 08:42
by oavina24
Thank you AL,
You have no idea how much this has helped.
But I have a question, is it absolutely necessary to "Increase the last point of 'Range' such that the whole experimental data are displayed". My data is pretty noisy at high q. What do you recommend?

Re: My data is pretty noisy at high q

Posted: 2020.02.20 11:56
by AL
If your experimental SAXS data at q > 8/Rg is just noise around zero (check by plotting your data on an absolute scale) there is not much point in using GASBOR. Just use DAMMIF.

Re: Help with GNOM/Gasbor

Posted: 2020.02.21 09:26
by oavina24
What i meant to say is that if its necessary to increase the whole range of data when calculating the pair distance distribution. Autognom gives the best quality estimate but does truncate the data.

Re: Help with GNOM/Gasbor

Posted: 2020.03.09 05:29
by oavina24

Re: bump

Posted: 2020.03.09 10:33
by AL
oavina24 wrote:
2020.02.21 09:26
What i meant to say is that if its necessary to increase the whole range of data when calculating the pair distance distribution.
Could you please rephrase your question?