SAXS Package Reference Page




What did I download with the SAXS data?

A folder containing 5 files.

1)    A file containing scalars extractable from SAXS – Target_Scalars.txt

2)    A “raw” SAXS profile 3 columns (q, Intensity, Error) – Target.dat

3)    A Fourier transform of the SAXS data 3 columns r, P(r), error – Target.pofr

4)    A PDB file from SAXS data outlining the shape of the molecule – Target.pdb

5)    A SITUS file from SAXS data outlining the shape of the molecule – Target.sit

6) A presentation in ".pdf" format summarizing all results 


More Information on SAXS data types from the .txt file


Experimentally Extracted Scalar Data types



Radius of Gyration from Gunier + Error


Angstroms (A)

Degree of Flexibility


Range (2 – 4) (Unfolded – Globular)

Experimental Mass


Daltons (Da)

Maximum Dimension + Error


Angstroms (A)

Radius of Cross-Section


Angstroms (A)

Experimental Volume


Cubic Angstroms (A3)

Radius of Gyration From P(r) Function


Angstroms (A)


1-Dimensional Data sets + Error (2x~500 matrix + error)


File Extension

SAXS profile



Electron Pair Distribution Function




Volumes Formats – SAXS shape cannot get correct “hand”

File Extension

Protein Data Bank Files


Situs Volume Maps



Relations between Structure and Extracted Parameters listed above

Rg is second moment of inertia of a structure – pymol has a calculator


Degree of Flexibility – The exponent of decay for a SAXS curve depends on flexibility.  q-2 for a flexible protein q-4 for a globular protein. A protein can have a flexibility anywhere in between. The absolute value of the exponent is listed.


Experimental Mass – Mass can be determined from a scattering profile as shown in

Rambo and Tainer Nature 496, 477-481 2013


Maximum Dimension – Largest atom to atom distance within a protein. Determined by defining where the P(r) function reaches zero at large r.


Radius of Cross-Section – Quantifies the second moment of inertia for a cross-section of the protein and is valuable when the cross-section is fairly constant – ideally rod like.


Volume – Calculated based on an integral of the scattering curve.


Radius of Gyration from P(r) function – Should be similar to that described above however this is calculated using the P(r) which in turn is calculated from the entire SAXS profile rather than a small region.


SAXS profile – can be calculated from PDB coordinates directly using FOXS calculator. Available as a script and web form


Electron Pair Distribution Function P(r) – A histogram of electron pair distances which is equivalent to mass pair distance. Relatively straightforward to calculate from a PDB, although water needs to be taken into account.




PDB based shape – produced from the algorithm GASBOR. Single shape representing the average shape in solution.


Situs Volume Map – Should match the GASBOR PDB but often easier to visualize over a target ribbon diagram as transparency can be used.

Quality of SAXS data - For CASP13 SAXS results are scored (Gold, Silver and Bronze) for the quality of results based on several quality judgement parameters and available controls. If samples were available in limited quantity all controls may not have been performed.


Challenges - When multimerization is known to occur or flexible regions appear to be factors they are noted as challenges.


What should I be aware of when I use SAXS data?

SAXS measures the entire particle in solution.  SAXS data will include electron contribution from disordered regions (90% of CASP12 targets) and will measure the particle as a multimer, if occurring in solution (50% of CASP12 targets). CASP results are scored for predicting specific portions of a target that may be only part of the protein or complex measured in SAXS. Use of SAXS data for prediction requires proper consideration of the contribution of the disordered regions or all parts of a multimer. 


It is possible to compare models using SAXS single value metrics (e.g. Rg), curves, or volumes.  If you use the reciprocal space curve for comparison, please note that the curve is always shown as a log plot because values can vary 103 over the range of q. If you use a subtractive comparison (e.g. Chi Square), the comparison will be weighted strongly towards the low q. 

More discussion of CASP12 and potential pitfalls were described in one of two April 2018 videoconferences (see link) and is described in

1: Ogorzalek TL, Hura GL, Belsom A, Burnett KH, Kryshtafovych A, Tainer JA, Rappsilber J, Tsutakawa SE, Fidelis K. Small angle X-ray scattering andcross-linking for data assisted protein structure prediction in CASP 12 with prospects for improved accuracy. Proteins. 2018 Mar;86 Suppl 1:202-214. doi: 10.1002/prot.25452.


Where can I get more Information?


Authored by Greg Hura and Susan Tsutakawa (LBL), on behalf of CASP