Previous: evalall: Integrate shoe boxes
Next: postrefine: Post-refine unit cell and detector position

Nanny: analyse data set

Usage:

  nanny [.y file name]

When the program starts, it will read in the ".y" file specified on the command line (if any), and display some statistics.

In the top panel three parameters are shown that are applied upon read-in of the data set.

"p" is the error factor according to McCandlish, Stout and Andrews. It is normally on the order of 0.02-0.04 depending on the instrument and the calibration. If strong reflections have high chi2 values, that might indicate that "p" needs to be raised, but it could also be an indication that absorption correction might significantly improve the data. If you are sure that the instrument error is the only source of error, the "Automatically optimize p" option under the "Options" menu will tune the value based on this criterion.
"min I/sigma" is the I/sigma level above which a reflection is considered as "observed". Normally this does not have any consequence for reflections with a positive observed intensity ("WEAK" reflections are a valid part of the data set), but this parameter also labels reflections with a significantly negative intensity as "NEGATIVE", and such reflections are usually rejected.
"reject" is the outlier detect criterion. If a reflection is more than the specified number of standard deviations away from the mean merged intensity for that h,k,l index, it is left out.

At the bottom left, a 2-pane notebook allows you to choose between the different statistics that the "any" program can generate.

The R-int tables will give the different R-factors for different classes of reflections. All of these tables will list:
- the total number of reflections used in the calculation (labeled "NRef" or "Reflections"),
- the number of unique data these represent (labeled "Uniq" or "Unique")
- the internal merge R ("Rsym")
- an improved merge R ("Rmeas")
- a chi-squared value ("Chi2") that expresses how well the differences between equivalent reflections are described by the standard deviations (chi squared is 1.0 if the standard deviations are good, higher values mean that the differences between intensities are larger than the standard deviation would indicate).
The "by pointgroup" table is special: it displays several merge results each including all reflections from the data set.
The listings can give samples of reflections with specific characteristics (e.g. the reflections with the highest chi2 values). Individual reflections can be "clicked" with the first mouse button. This will bring up the makextalevc GUI showing the integration of that particular reflection.
The number of lines in these listings is limited. If any listing is cut short at the bottom, this is clearly indicated. Normally the first 200 lines (the default length) are sufficient to get a reasonable feel for the diagnostics, but if you want to get more output, the limit can be set using the "Maximum listing length" option under the "Options" Pull down menu.

Under the "Options" menu one further choice is important: "Change pointgroup" will ask you for a new laue group to perform data merging. This can be useful if the lattice has pseudo-symmetry.

Filters menu

On the filters menu you will find options to select subsets of the data that you want to use.

The "Reflection flag selections" option allows you to choose to use reflections with certain error flags assigned by the integration process. If you select this option, each of the flags that occurs in your data set will be shown and you can choose for each of these whether you want to include them in your data set. By default, only good reflections are allowed, plus any reflections marked "weak" and/or "overlapsplit". A help balloon will show up explaining each of the flags when you hover over them with the mouse (only if balloon help is switched on).
The "scan selections" option allows you to eliminate all data from certain scans selectively from the data set.
The "Theta limits" or "Resolution limits" option allows you to eliminate high-order or low-order reflections from your data set.

Output

Once you are happy with the selected subset of data:

it can be written to a Shelx "HKLF 4" type file using the "Write nanny.hkl" option in the "File" menu.
If you are dealing with a twin, and have done integrations of both twin components, you can also write a "HKLF 5" type file using the "Write nanny.hklf5" option in the "File" menu. The output file produced can not be used to enter the maXus program, but it can be used for twin refinement in SHELXL97.
If you are dealing with an incommensurately modulated structure described with a 4-dimensional HKLM lattice, you can use the "Write jana.hkl" option in the "File" menu to export your data to JANA98 or JANA2000.
If you want to scale your data, you can use the "Write shelx.sad" option in the "File" menu to export your data to the "sadabs" program from the shelxtl package (only versions newer than july 2002 of sadabs support evalccd).

Previous: evalall: Integrate shoe boxes
Next: postrefine: Post-refine unit cell and detector position