· Delarue Home
· AquaSAXS
  · Introduction
  · Flowchart
  · News...
  · SAXS profile (auto)
  · SAXS profile (expert)
  · Sequential fit
  · Ensemble fit
· References
· Links
· BioIsis test set
· Job queue status

We detail here what the program actually does, and the available options.

Input PDB/PQR file editing

One of the first steps in the program is reading and editing the input PDB/PQR file. Both formats are well interpreted, although the PQR format is necessary if you want to use the "Solvent-grid" approach without providing your own map.
By default, the water molecules are not considered. That is, any line in the PDB/PQR file with the residue type "HOH" is not stored in memory.
The user has the possibility to tell the program to consider only a subset of the file, by telling which chains must be read. This is useful, for example, when one wants to discard some ligands unlikely to be present in the sample solution used for SAXS experiments.

Atomic types

The program considers Hydrogen atoms implicitely. If they are present in the PDB/PQR file, they are not read. Nevertheless, Carbons with zero, one, two or three bound Hydrogens consitute distinct scatterers, i.e. have distinct atomic types.
When the PDB/PQR file is read, the atom and residue names are compared to a library (taken from the Protein Data Bank) in order to assign the atomic type.

If you want to check that the atoms in your PDB will be recognized, parse the residue name here:

Note that the program will list in the output logfile all the non-recognized atoms.

We also let the possibility for the user to define its own atomic types and residue equivalence tables.

How to define a new residue type?
Here is an example of file that the user can give to the program in order to define the residue type "DDM". Note that if the new residue name already exists in the program library, it will overwrite it.
This file can define several residue types, provided they are separated by "...".
The residue's atoms are listed, one atom per line. The first column is the atom 4-letter name as found in the input PDB/PQR file. The second one is its type: e.g. "CH2" if the atom is a Carbon with 2 bound Hydrogens.
Below we explain how to define the atomic type "XXX", that can be used in your new residue type definition.

How to define a new atomic type?
Here is an example of file that the user can give to the program in order to define the atomic type "XXX". This file can define several atomic types, provided they are separated by "...".
The five first lines define the atom's form factor as the sum of a constant (G5) and four Gaussians (see file for explanation). The sixth line (starting with "Xv") gives the solvent-excluded atomic volume, and the seventh, the corresponding radius.

Solvation method

AquaSAXS offer two ways to account for the hydration layer's contribution to the SAXS profile of a solute model. These options are called "Solvent-map" or "Surface-accessible". The latter is here for comparison to the former, and is based on the method proposed in the FoXS web server.

  • The "Solvent-map" method:
    • Recently developed methods, such as 3D-RISM or AquaSol, allow one to compute a solvent density map from a given PDB model. One major novelty of AquaSAXS, in comparison to already existing programs, resides in its ability to take advantage of these solvent density maps when computing the theoretical SAXS profile of the same PDB model.
      A solvent density map is basically a cubic grid centered on the solute, with a given number of points in each direction, each separated by a given distance (hence defining the map resolution), with a given density value at each point. Basically, in such maps, one would expect a density of 0 at a solute atom's location, and 1 in the bulk. At the boundary between the solute and bulk's region, the density varies.

      In a first approximation, one would define a layer around the solute within which all grid points would get the same value: this is the solvation model used by CRYSOL. Now, what 3D-RISM and AquaSol do is computing the solvent density value at each grid point on the basis of the physical interactions between the objects in the system.
      In the case of AquaSol, we showed that the resulting water distribution profile was in good agreement with experimentally observed ones, and respected the chemical nature of the solute's exposed atoms.

      In AquaSAXS, whether the user provides its own map or lets the program compute one (using AquaSol in a default mode), the program reads it (provided the map in the CNS format), stores the points and their density value that are outside the solute, subtract one to the value to get the excess density value, and compute the Fourier Transform of the resulting map. The result is the hydration layer's structure factor used in the SAXS intensity formula.

      3D-RISM has been recently implemented in the AMBER tools package.
      AquaSol exists as a web server open to all users (it is also freely available from us). To run the online-version of AquaSol, click here.

  • The "Surface-accessible" method:
    • In this method, every solute's atom is assigned an additional term to its form factor to account for the hydration layer's contribution. This term is proportional to water's form factor, the proportion being given by the fraction of solvent-accessible-surface of the atom.

      This solvation model is directly taken from the FoXS method, and is only here for comparison to the "Solvent-map"'s solvation model. Although less physically grounded, this method should generally be faster since the summation is made over the solute's atom only.

    Fit to experimental profile

  • Practical considerations
    • In order to fit the theoretical SAXS profile to an experimental one, the user must provide an ASCII file containing the experimental data.
      Every line starting with a number from 0 to 9 or a dot "." will be considered as containing data.
      The experimental profile file must contain at least 2 columns: the first column with the value of the wave-vector norm q, the second with the measured intensity value. The third column is optional, but is dedicated for measurement errors.
      If this column is empty, we will define it as the difference between the value at q(n) and the value averaged from q(n-1) and q(n+1).
      Please note that columns must be separated by spaces, not tabs.
      Please also note that the presence of control characters in the ASCII file might lead the program to fail. Please make sure that only real numbers are present in the lines containing data.
  • Computational considerations
    • We need here to display a few maths...
      The calculated intensity ΔI(q) scattered by the solute is the result of the following spherical averaging over all directions of the wave vector q=(q,Ω), where A is the system's excess structure factor:

      A is the result of three contributions: Fsolute, Fsev, Fhs, due, resp., to the solute, the solvent-excluded-volume and the hydration shell.
      The two latter are weighted by the solvent bulk density ρw and two additional parameters are used to adjust the relative contributions: C1 and C2.
      The user is free to set the values of ρw, C1 and C2. Though, he can let the program find the optimal values of C1 and C2, given ρw, if an experimental profile is provided. In that case, the following quantity χ is minimized:

      To minimize χ a given range of values is spanned for C1 and C2. For each pair, a least-square minimization is performed to get the scaling factor C.

      The pair that yields the lowest χ is kept, and the corresponding profile is retrieved by the user.

      The expansion factor G(q,C1) is a gaussian in q-space defined by the following equation:

      where rm is the average atomic radius.

      Marc Delarue http://lorentz.dynstr.pasteur.fr
    Page last modified 14:57 April 27, 2011.