Large dynamic range small-angle diffractometer

Short instructions for the use of SASHA and ASSA

Below you will find short instructions how to use beta-versions of SASHA and ASSA by Dmitri Svergun. A comprehensive help for SASHA and ASSA is being prepared. Instructions for CRYSOL are here (html - 34 Ki). The information below corresponds to the file /hosts/lass2/d1/lss/svergun/Assa/readme.

At the ILL, use the commands assa and sasha. For ASSA, the directory Assa needs to be present in the user's home directory. You may copy it from ~may (/hosts/d22sgi/d1/lss/may).

The SASHA executable files are:


2d plotting
Flm coefficients of an ellipsoid
partial amplitudes from Flm coefficients
required by DSPLOT
shape determination

they should be in a directory on the path. They were all compiled under IRIX 6.4.
If gnuplot is running on your machine, do not use the supplied gnuplot files
(check with your system administrator).


The ASSA executable files are:


2D plotting, used by ASSA
intensity from two bodies, used by ASSA
3D graphics
intensity from Flm coefficients, used by ASSA
shape determination
surface from Flm coefficients, used by ASSA

NOTE: trans6, not trans3 !

they should be in the directory $HOME/Assa, which should be
added to the path. They were also compiled under IRIX 6.4. Also
in the directory $HOME/Assa should be the following auxiliary files:

NOTE that ASSA will not work properly unless all the necessary files
are stored into this directory.


Test files:

fixed.pdb, fixed.flm, fixed.alm

moved.pdb, moved.flm, moved.alm

PDB structure, envelope and partial amplitudes of lysozyme

PDB structure, envelope and partial amplitudes of lysozyme (shifted and rotated) comp1.dat scattering intensity from a complex (fixed+moved)

Environment variables:

setenv GNUTERM x11
setenv GNUPRINT postscript

Working with SASHA

Sasha normally works in the user mode with default answers.
The program may stop at the very beginning if it is unable
to compute the Rg using Guinier approximation. It is made
deliberately: is one cannot compute Rg, one should not try
shape determination; check your data if this happens.

For typical experimental data, default (L=4) resolution
can be taken. If you apply symmetry restrictions, those
reduce the number of free parameters and may allow to use higher harmonics.

The output fit is normally placed onto <outname>.fim; the resulting

shape onto <outname>.flm

If your structure is a dimer, you could e.g. treat it as
a single body with the P2 symmetry. In this case you enter
the experimental data as MONOMER curve, and the outputs
will be as usual, *.fim and *.flm.

Another option is to look for the shape of the monomer and
for the distance between the monomers. Then enter the curve as
DIMER curve (answer just <cr> to MONOMER). The
outputs will be *.flm (monomer shape; the distance is written at the
bottom of the file; see ASSA below how to display a dimer), and *.fiz
(fit to the data).

If you have both scattering curves (from dimer and monomer)
you can enter both of them; the program will try to simultaneously
fit them). The output files will then be *.flm, *.fim, *.fiz.

The fits can be viewed by typing dsplot <outname>.fim (or fiz),
the shapes with ASSA (see below).

Working with ASSA

ASSA loads the atomic structures (*.pdb files) and the surfaces
(*.sld files) defined by sets of multipole coefficients (*.flm files).
If you load an *.sld file (File -> Load), it is loaded
as is; if you load an *.flm file, ASSA invokes TRANS3 to (re)compute
the surface. If ASSA crashes when loading an *.flm file, please

check whether TRANS3 is in the $HOME/Assa directory and that

it works properly (e.g. type trans3 <filename> in the current
directory: *.sld file should apppear).

To display a dimer with a two-fold axis (e.g. that from SASHA's
output), just put the value of the distance between the monomers
into 'Dimer dist' while loading the monomer: the second monomer
is loaded automatically and the entire dimer is considered as a
single body.

If you want to compare *.pdb and *.sld models of the same object,
be sure to place the *.pdb model to the center of coordinates
(Load -> Reset_to_0). The *.sld model will be centered automatically,
unless the other position is specified. You may use transparency
(Options->General->Transparency), and rotate the whole picture
to get the best view (Select the ROOT by clicking outside the
models and use the transformation sliders).

Interactive modelling with ASSA

Suppose you have measured the scattering curve (ab.dat) from a
complex of two domains, A and B, and their atomic models are
available (say, a.pdb and b.pdb). To fit the experimental scattering
curve, the following should be done:

  1. Compute the scattering amplitudes from the two structures
    using CRYSOL (up to L=10). If the scattering from them was not measured
    separately, use default contrast in the shell and default volume of the atomic group.

    NB: the number of points, maximum order of harmonics
    and the maximum scattering vector value MUST be EXACTLY the same for the both structures.
    The maximum scattering vector should be
    larger than that in the experimental curve ab.dat.
    Note the definitions of the angular units (1-4), see instructions for Crysol.
  2. Rename the amplitude files you obtain (e.g., a00.alm and b00.alm
    to a.alm and b.alm, respectively).
  3. Start ASSA and load the two structures a.pdb and b.pdb (File -> Load)
  4. Start the ALM22INT program (Command -> Complex intensity), specify
    the data file, its units, angular portion of the file you would
    like to wit (in fractions from the expermental smax). Select
    the bodies (e.g. by typing or by clicking on them). After moving
    to a desired positions, click Start: the fit will be displayed.
    You may change the position and click Start again as many times
    as you like (the 2D plot windows must, however, be killed
    manually at the moment). The current fit is saved onto the output
    file specified by the user.
    If the pdb structures have different contrasts, this can be taken
    into account by the pre-multipliers.


You can check the modelling in Assa by loading the bodies fixed.flm
(or fixed.pdb; they both point to fixed.alm) and moved.alm(or .pdb).
Invoke "Complex intensity" and give comp1.dat as an input data file.
As loaded, the two bodies yield Chi=1.6.

Graphic output

For 2D pictures produced by DSPLOT, a postscript file
/tmp/gnu-capture is created which contains
the plot; this file is rewritten after the next picture is plotted!

At the moment, there is no build-in screen dump for ASSA; one can use
any of screen grabbers available for SGI (e.g. that in mediarecorder),
and then use "imgview" to edit and/or print the image.