The Spotter Elves analyze the output of scala and truncate, or scalepack. They examine these logs for expected systematic absences, and rejected ("outlier") spots. These spots should really be inspected visually (that is, by you), but seldom are. This is largely because it is far too time-consuming to sift through the massive amount of data typically collected these days (~1000 frames). This is where Spotter Elves come in. They take the list of expected absences and rejects, and then look into the processed data file (either a raw mtz or the denzo *.x files), for the x,y detector coordinates and frame number of these spots. They then go to the raw data images (which must still be on disk), and extract the raw pixels from each region of interest. The same region is grabbed from adjacent frames, for rocking-curve context. All these images are then assembled into either an SGI movie (on SGIs only) or an animated GIF (using ImageMagick). If ImageMagick is available, the hkl and frame filename will be printed directly onto the graphics image, uniquely identifying it. If it is not available, a text file called spots.list tells you which frame of the movie corresponds to which hkl on which image.
Therefore, using Spotter Elves, you can review the critical spots from thousands of frames in just a few minutes.
To do all this, Spotter Elves make use of the adsc2pgm and osc2pgm programs. At the moment, only ADSC Quantum 4 and Raxis II and IV type images can be used. If anyone out there wants to write a mar2pgm, let me know, I havn't figured out how to read the compressed mar images. Apropriate binaries of these programs should be located in your execution path for Spotter Elves to run properly. However, in case they aren't, Spotter Elves carry uuencoded copies of these programs for sgi and Intel linux within the Spotter script, as well as the c source (in case you have a compiler). If all else fails, they will try to use an awk version of these programs, but it is really slow. Unless you are on an SGI, Spotter Elves will also not be able to run without ImageMagick. Fortunately, ImageMagick is free, and pre-compiled binaries are available for most computer systems.
In the future, Spotter Elves will be able to take an arbitrary list of hkl coordinates, optionally symmetry-expand them, and show you a movie of whatever you like.
Although your crystal's space group is one of the first things you seek to determine about it, you never really know which one it is until your structure is finally solved, and you can see the hand of the alpha helicies. The reason for this is the loss of translation information in the diffraction pattern, which makes it impossible to determine the hand of your space group (P41 vs P43, etc), and nearly impossible to differentiate between rotation and screw operations (P2 vs P21, etc.). The hand problem can only be resolved during phasing, but the screw vs rotation problem is usually elucidated by looking at systematic absences.
Spotter Elves take advantage of the moviefy script, which you can download from here too. Moviefy is good for more than just spot movies. You can also make movies of large regions of the detector area, and watch the lunes of your reciprocal lattoce planes whirl by.
So, how do you use Spotter Elves? Run the script with the log file output from your scala/truncate or scalepack run on their command line:
unix% Spotter Elves logs/scala_truncate.log /data/jamesh/frames/frame_001.img
Or, if you have denzo data:
unix% Spotter Elves scalepack_out.log
Spotter Elves are smart enough to figure out almost everything from just the log file, but, if you have moved files around, you should tell them the real location of your first frame on the command line. Note that mosflm does not (necessarily) record the location of the original raw image data files in its output, so you should make sure you list them there.
If you just have a scala run, with no truncate, you will not see systematic absences.
Spotter Elves alpha-version Caveats
As of the current alpha release:
Back to the Elves Manual Table of Contents.