Extract and Integrate Reflections
Syntax: iplt_diff extract [-f] [-i MASK] [-v VERB]
This is the major workhorse of the individual image diffraction processing. It extracts various parameters for each predicted reflection based a lattice, which comes either from a previous manual lattice verification in form of a VerifiedLattice, or a RefinedLattice from a previous extraction run.
Based on the lattice, a 2D gaussian fit is performed at each possible lattice point location, using LatticeGaussianExtract. The lattice with is then then refined, based on the centers of the 'well behaved' gaussian fits, resulting in slightly updated positions for the predicted lattice point positions.
The reflections are exported to the _ext.mtz file, with each lattice point characterized by values from the gaussian fit.
Syntax: iplt_diff integrate [-f] [-i MASK] [-v VERB]
A peak integration algorithm is now run for each refined reflection position, using PeakIntegration, and a subsequent filtering again removes the clearly non-usable reflections. This peak integration takes the background tilt (Px and Py) from the gaussian fit into account, but not the absolut offset (C).
As a last step, the intercalated background values are determined, based on the average of the four points halfway to the neighbouring lattice positions.
Finally, the reflections are exported to the _int.mtz file, with each lattice point characterized by values from the gaussian fit, the peak integration, and the intercalated background. Several of these values are used in the subsequent tilt step to calculate the intensity and the sigma for each reflection.
Two debug images are generated, called tmp_integrate1.tif and 'tmp_integrate2.tif`. Both are assembled from the individual profile boxes that are used for the gaussian fitting and peak integration, but the former contains the profiles without and the latter with background correction.
Parameters in project.xml
Under the group LatticeExtract: (also used for the integration step)
- BoxSize (integer): half-size of initial, inner, square profile
- BoxGrowAmount (integer): additional grow-steps to take during peak integration
- BackgroundRimSize (integer): width of background rim surrounding inner profile region
The actual subimage size used in the gaussian fit is (BoxSize+BoxGrowAmount+BackgroundRimSize)*2+1, the one in the peak integration is (BoxSize+N+BackgroundRimSize)*2+1, where N goes from 0 to BoxGrowAmount.
<DiffProcessing> <LatticeExtract> <BoxSize value="6" type="int"/> <BoxGrowAmount value="1" type="int"/> <BackgroundRimSize value="2" type="int"/> </LatticeExtract> </DiffProcessing>
Results in info.xml
Under the group DiffProcessingResults/DiffExtract/
- RefinedLattice: the refined lattice, which will be re-used upon a second extraction step
