2.15 *******************
* CMBANO WRITE-UP *
*******************
CMBANO is a program to merge native data with derivative anomalous
scattering data. It is used to prepare a "merged" file for use in
PHASIT to phase by derivative anomalous scattering, to make an input
file for TOPDEL to create Bijvoet difference Patterson coeficients, to
make input files for MRGBDF for Bijvoet difference Fouriers and to get
input files for GREF for refining heavy atom sites against anomalous
scattering differences. The program is interactive and prompts for the
names of the input and output files, and whether or not one wants to
include anisotropic scaling corrections. The program matches up all
derivative Bijvoet pairs with the corresponding native data, and
scales the derivative data to the native. The reflections need not be
indexed identically in both input files as symmetry information is
used to match up the data. Each input data set however, should contain
only unique reflections. The data sets are initially scaled by
computing a relative Wilson plot and applying the scale and thermal
corrections derived from it. The user is then asked whether any
additional non-Wilson scaling should be done. If it is, then the
user is asked whether anisotropic or local scaling should be done.
If anisotropic scaling is requested, the reciprocal lattice vectors
are orthogonalized and the elements of a symmetric 3x3 scaling tensor
are refined by two cycles of least squares. The anisotropic scaling is
then applied to all of the derivative reflections, and the tensor
elements are printed out. For an isotropic distribution the diagonal
elements should be 1.0 and the off diagonal elements zero. Thus
deviations from these quantities indicate the degree of anisotropy. If
local scaling is requested, then a scale factor for each reflection is
determined by a least squares fit of the F's for all neighboring
reflections within a given sphere radius to the corresponding native
F's, neglecting the central reflection to be scaled. For each
reflection the sphere radius is initially chosen to encompass about
125 reflections, and the derived scale factor is accepted if at least
80 neighbors are found. If needed, the sphere radius will be
incrementally adjusted until either a preset maximum is reached, or
80 neighbors are found. If the maximum is reached, then the scale
factor will still be accepted if 40 neighbors are found. If not,
the program will stop and indicate that the data set is too sparse
for meaningful local scaling. The mean and minimum number of
neighbors used is then listed. For both anisotropic and local
scaling, the minimum and maximum scale factors that were applied are
listed. Note that the additional scaling (anisotropic or local), if
invoked, is applied ONLY TO SCALE THE DERIVATIVE DATA TO THE NATIVE,
and NOT to scale F+ to F- within the derivative. Thus one may still
want to apply some type of additional scaling to the F+,F- values
prior to running this program.
***** FILES *****
The native data file must contain Miller indices and the corresponding
reflection data values. The derivative data file must contain Miller
indices and the corresponding reflection data including Bijvoet pairs.
The files however, can be one of three types. If the file name ends
with ".MU" or ".mu", then it is assumed to be a "MULIST" i.e. a file
generated by program MAKEMU (in the XENGEN system) or by program
FBSCALE. In that case each record is assumed to contain
H, K, L, RES, F, Sig(F), F+, Sig(F+), F-, Sig(F-), Iflag
in format (3I4, 1X, F6.4, 6(1X, F8.2) 1X, I2 ). For the native file
only the indices, F and Sig(F) are used. For the derivative file only
the indices, F+, Sig(F+), F-, and Sig(F-) are used. If "Iflag" is
present in the derivative file, then it will be used to screen for
viable anomalous scattering data, and one can consult the XENGEN
documentation for the meaning of the Iflag variable. If it is absent,
then only reflections with both F+ and F- values greater than zero
will be used. (This criteria also applied even if Iflag is present).
If the filename ends with ".SCA" or ".sca", then a SCALEPACK
file is assumed. After a variable number of header records
(see the FILE FORMATS section), reflection records follow and
contain
H, K, L, I+, sig(I+), I-, sig(I-)
in format (3I4, 4F8.1)
Note the use of intensities rather than F's. The last two items
in each record may be omitted FOR THE NATIVE SET. If present in the
native file they would be used only if I+ was not measured. For the
derivative file in general all quantities should be present in each
record as if either member of the Bijvoet pair is missing the
reflection will not be used. Only reflection with all F's greater
than zero will be used.
If the filename does not end in ".MU", ".mu", ".SCA" or ".sca"
each record is assumed to contain
H, K, L, F, Sig(F)
if it is the native file and
H, K, L, F+, Sig(F+), F-, Sig(F-)
if it is the derivative file. The file is then read in free format
i.e. each item must be separated by at least one space or a comma. The
indices must be INTEGERS and all F and Sig values REALS. Reflections
are accepted only if the F (or F+ AND F-, for derivative data) is/are
greater than zero.
It is absolutely ESSENTIAL that only valid measurements of BOTH
F+ and F- are used by the program, and the screening criteria above
is designed to insure that. However, depending on the source of the
data it is still possible for invalid reflections to be used. For
example, some data reduction programs output only the mean F and
del ano, and the user then writes a small program to convert this
to the individual F+, F- values. If a del ano of zero is encountered,
it MAY mean that one of the F's (either F+ or F-) WAS NEVER MEASURED!
Yet the conversion program might output this as a Bijvoet pair with
a Bijvoet difference of zero! Likewise some data reduction programs,
even if they output individual F+ and F- values, actually set one
of them equal to the other if only one was measured. This again leads
to erroneous Bijvoet differences. Know what your data reduction
program is doing, and be very wary of any non-centric reflections
with F+ EXACTLY equal to F-, or have del ano equal to zero!
The output file contains records with
H, K, L, FP, Sig(FP), FPH+, Sig(FPH+), FPH-, Sig(FPH-)
in format ( 3I4, 6F10.2). The last 4 quantities are rescaled to match
the native data set. This file is suitable for input to PHASIT using
the ISOFLG=2 option, or for input to MRGBDF to prepare coefficients
for Bijvoet difference or cross Bijvoet difference Fouriers.