Step #2: Define the volume within which all points are at least 1.2 Angstrom away from the
nearest van der Waals surface. This volume is obtained in a way similar to that in step #1 but
now with atom radii being the van der Waals radii increased by 1.2 Angstrom. The volume
that is found in this way will be called the Ohashi volume.
Step #3: Extend the volume obtained in step #2 with all points that are within 1.2 Angstrom
from its bounding surface.
STEP #1 – EXCLUDE VOLUME INSIDE THE
VAN DER WAALS SPHERE
DEFINE SOLVENT ACCESSIBLE VOID
DEFINE SOLVENT ACCESSIBLE VOID
STEP # 2 – EXCLUDE AN ACCESS RADIAL VOLUME
TO FIND THE LOCATION OF
ATOMS WITH THEIR
CENTRE AT LEAST 1.2 ANGSTROM AWAY
DEFINE SOLVENT ACCESSIBLE VOID
STEP # 3 – EXTEND INNER VOLUME WITH POINTS WITHIN
1.2 ANGSTROM FROM ITS OUTER BOUNDS
Fig. 5.2.1-1
.
Cartoons illustrating the three stages of the identification of the solvent
accessible volume.
5.2.2 The Numerical Implementation
The numerical implementation in PLATON of the solvent region model defined in
Section
5.2.1 is relatively compute intense. The calculations are based on a grid with a distance
between grid points of in the order of 0.2 Angstrom. The exact value depends on the
requirement to have an integral number of grid steps in each of the three dimensions.
The algorithm may be summarized as follows.
1.
The unit cell is filled with atoms of the (symmetry expanded to P1) structural model
with van der Waals radii assigned to each atom involved. The default van der Waals
Radii can be customized with the
SET VDWR ELTYPE radius (ELTYPE radius ..
) instruction [e.g.
SET VDWR C 1.7 H 1.3 O 1.8].
2.
A grid search (with approximately 0.2 Angstrom grid steps is set up to generate a list
of all grid points (list #1) in the unit cell with the property to be at a minimum
distance of 1.2 Angstrom from the nearest van der Waals surface.
3. The list generated under 2) is used to grow lists of grid points (possibly supplemented
with grid points within 1.2 Angstrom around list #1 points) constituting (isolated)
solvent accessible areas.
4. For each set of 'connected grid point sets' a number of quantities are calculated.
1. The center of gravity of the void.
2. The volume of the void and the (Ohashi) volume of grid points where the
centers of the atoms of the solvent can reside.
3. The second moment of the distribution (The center of gravity can be seen as a
first moment). The corresponding properties of the second moment (ellipsoid)
can be calculated via the eigenvalue/eigenvector algorithm. The shape of the
ellipsoid can be guessed from the square-root of the eigenvalues: a sphere will
give three equal values.
5. For each void in the structure a list of shortest distances of centre-of-gravity of the
void to atoms surrounding the void is calculated. Short contacts to potential H-bond
donors/acceptors may point to solvents with donor/acceptor properties.
5.4
– Informal Theory of the SQUEEZE Procedure
Fig. 5.4-1. Determination of the contribution of the solvent region to the structure factor.
The essence of the procedure is that the total electron density is split up into a contribution
of the ordered model (which is approximated generally as the sum over atomic distributions
leading to the normally used expression for the structure factor) and a contribution of the
disordered solvent. The latter is approximated by the densities at grid points. That density is
transformed into a structure factor contribution with a discrete Fourier transform. The
determination of the disordered solvent contribution is achieved by masking out that density
in a difference density Fourier map. That procedure is necessarily iterative because of the
well known problem that the initial difference map has to be calculated with the phase
assigned to F(obs) to be the same as F(calc). Eventually, the solvent contribution will have a
phase that will differ from the phase of the model F(calc). A solvent free F(obs)' is obtained
by taking the complex difference of F(obs), phased with the phase of the total F(calc) (i.e.
model + solvent) and F(Solvent). The number of electrons in the solvent region is estimated
by proper addition of the density found at grid points in the solvent region. Missing strong
or erroneous low order reflections may make the latter count useless.
5.5 - Recommended SQUEEZE Procedure
Step 1: Refine a discrete atom model including hydrogen atoms with SHELXL.
Step 2: Delete (when applicable) all 'atoms' used to tentatively model the disordered region.
Step 3: Do a PLATON/SQUEEZE run with
.res from
Step 2 and
.hkl used in
Step 1. A file
.hkp is produced.
Step 4: Copy
.res from
Step 2 to
.ins and
.hkp to
.hkl from
Step 3 into a new directory.
Step 5: Refine with SHELXL in the directory created in
Step 4.
Step 6: Analyze the result and optionally repeat
Step 3 with the
.res from
Step 5 and the
.hkl used in
Step 1.
Step 7: Do a final 'CALC FCF' with PLATON to get a proper
.fcf (with the original
observed intensity and the calculated intensity based on both the model and the solvent
contributions) with the
.res and
.hkl from
Step 5. The result in in
.hkp.
Step 8: Rename
.hkp into .
fcf
Step 9: Append the SQUEEZE produced info in file
.sqf to the
.cif from
Step 5.
5.6 - Concluding Remarks
Voids containing disordered solvents are often located at special positions. The solvents that
occupy those sites generally have the volume needed to fill the space between the main
molecules of interest but not the point group symmetry compatible with the site symmetry
(e.g. a THF in a 3-bar site). Other ‘popular’ disorder sites are 3, 4 and 6 fold axes.
Checking for solvent accessible voids is also done as part of the IUCr CHECKCIF structure
validation procedures. In practice it is found that voids and their contents are not always
clear and thus easily missed. Density plateaus or ridges might evade peak search algorithms.
The ultimate tool to inspect void regions is the calculation of contoured difference maps,
either in terms of 2D sections or rotatable 3D maps (see e.g. Tooke & Spek, 2003).
More details can be found in
Section 1.3.3.3.