Cppxfel Post-refinement: Difference between revisions
No edit summary |
No edit summary |
||
Line 125: | Line 125: | ||
img-shot-s01-20130316165021253_0.mtz 0.991536 0.114612 0.787729 217 0 1.46545 0.0013 -0.0109289 -0.0157625 8.25238e-05 1.5 105.783 105.783 105.783 | img-shot-s01-20130316165021253_0.mtz 0.991536 0.114612 0.787729 217 0 1.46545 0.0013 -0.0109289 -0.0157625 8.25238e-05 1.5 105.783 105.783 105.783 | ||
</pre> | </pre> | ||
== Next steps == | |||
After one is happy with the merging statistics for post-refinement, it is time to make a [[Cppxfel Final Merge | final merge]], with recalculation of sigma values. |
Latest revision as of 22:21, 24 November 2015
Post-refinement by cppxfel of integrated intensities occurs after initial orientation matrix refinement. This is the penultimate stage of producing a viable MTZ file for structure solution. This is also the point at which indexing ambiguities are broken (note that only two-fold indexing ambiguities can be broken at the moment without an external reference MTZ).
Contents of refine.txt
This stage is controlled by the input file refine.txt
generated by cppxfel.input_gen
and uses a number of files generated at the previous stage. The contents of this file:
cat refine.txt
ORIENTATION_MATRIX_LIST refine-orientations.dat MATRIX_LIST_VERSION 2.0 NEW_MATRIX_LIST refined.dat PARTIALITY_CUTOFF 0.2 COMMANDS REFINE_PARTIALITY
Executing refine.txt
This can be executed using the line
cppxfel.run -i refine.txt
This begins by generating an initial, merged MTZ file using an initial guess at the parameters of the image. For space group I23, this requires the indexing ambiguity to be broken, which can be followed by the values of the fx:
lines during L-BFGS refinement. The processing statistics for the first merge are then displayed.
This is followed by a number of rounds of post-refinement (approximately 6). The major processing statistics for each image can be followed as they are produced, followed by a large output of all the parameters used in refinement after each cycle. The processing statistics for each subsequent merge are then displayed until refinement ends.
Merging statistics output
The initial images are merged after attempting to break the indexing ambiguity and the Rsplit and CC1/2 are output from the two halves of the data set. An example of a good Rsplit and CC1/2 is shown below. The merging statistics are measured in resolution shells specified by the first two columns, followed by the value and the number of reflections which contributed to that calculation.
N: === R split === N: inf 4.31119 0.0880855 971 2 N: 4.31119 3.4218 0.0860462 949 2 N: 3.4218 2.98922 0.142905 1022 2 N: 2.98922 2.71588 0.197199 1017 2 N: 2.71588 2.5212 0.20902 1119 2 N: 2.5212 2.37254 0.255386 1116 2 N: 2.37254 2.25371 0.280964 1174 2 N: 2.25371 2.1556 0.285217 1157 2 N: 2.1556 2.07261 0.298872 1163 2 N: 2.07261 2.00108 0.334966 1228 2 N: 2.00108 1.9385 0.372736 1160 2 N: 1.9385 1.88309 0.45132 1151 2 N: 1.88309 1.83351 0.502127 1035 2 N: 1.83351 1.78877 0.602383 907 2 N: 1.78877 1.7481 0.628576 746 2 N: 1.7481 1.7109 0.768211 574 2 N: 1.7109 1.67667 0.848593 351 2 N: 1.67667 1.64503 0.898865 257 2 N: 1.64503 1.61565 0.882403 115 2 N: 1.61565 1.58826 0.989214 22 2 N: *** Overall *** N: 0 1.58826 0.159739 17234 2 N: === CC half === N: inf 4.31119 0.985825 971 0 N: 4.31119 3.4218 0.982884 949 0 N: 3.4218 2.98922 0.963 1022 0 N: 2.98922 2.71588 0.917537 1018 0 N: 2.71588 2.5212 0.90104 1121 0 N: 2.5212 2.37254 0.825298 1117 0 N: 2.37254 2.25371 0.849232 1175 0 N: 2.25371 2.1556 0.846396 1158 0 N: 2.1556 2.07261 0.856127 1164 0 N: 2.07261 2.00108 0.811718 1229 0 N: 2.00108 1.9385 0.775128 1162 0 N: 1.9385 1.88309 0.659263 1155 0 N: 1.88309 1.83351 0.586403 1041 0 N: 1.83351 1.78877 0.478341 914 0 N: 1.78877 1.7481 0.35525 761 0 N: 1.7481 1.7109 0.223638 596 0 N: 1.7109 1.67667 0.16558 374 0 N: 1.67667 1.64503 0.0562969 280 0 N: 1.64503 1.61565 0 127 0 N: 1.61565 1.58826 0 23 0 N: *** Overall *** N: 0 1.58826 0.982438 17357 0
Real-time processing statistics
During refinement the important statistics on individual images are reported as they are refined. The first column specifies the name of the MTZ file and the second column specifies the target function for minimisation. The third and fourth columns show the correlation coefficient and R factor with the reference MTZ, respectively. The fifth column is the correlation of the partialities of individual reflections with the estimated 'true' partiality by comparing against the reference data set, and should be as high as possible. The sixth column is the current Rmerge across all images calculated only for the reflections shared with the refined image. The seventh column is the B factor (unrefined in this case), and the final column is the number of accepted reflections after post-refinement due to the setting of PARTIALITY_CUTOFF
.
Refining image 124 img-shot-s00-20130316165200803_0.mtz img-shot-s00-20130316165051970_0.mtz rfactor 0.550657 0.600949 0.401253 0.320187 0 171 Refining image 71 img-shot-s01-20130316165048878_0.mtz img-shot-s00-20130316165008012_0.mtz rfactor 0.991731 0.111634 0.819409 0.324124 0 208 Refining image 69 img-shot-s01-20130316165031170_0.mtz img-shot-s00-20130316165323342_0.mtz rfactor 0.973724 0.223516 0 0.312648 0 185 Refining image 68 img-shot-s00-20130316165111763_0.mtz img-shot-s00-20130316165002470_0.mtz rfactor 0.825142 0.575939 0.277102 0.318178 0 191 Refining image 64 img-shot-s01-20130316165033837_1.mtz img-shot-s01-20130316165129338_0.mtz rfactor 0.987785 0.113023 0.7968 0.323475 0 198 Refining image 91 img-shot-s00-20130316165111763_1.mtz img-shot-s00-20130316165049012_0.mtz rfactor 0.988474 0.120959 0.548244 0.317826 0 208 Refining image 102 img-shot-s00-20130316165037095_0.mtz img-shot-s00-20130316164958596_1.mtz rfactor 0.989014 0.143094 0.7369 0.306314 0 189 Refining image 75 img-shot-s01-20130316165017753_1.mtz img-shot-s01-20130316165126505_0.mtz rfactor 0.977392 0.186307 0.633476 0.32707 0 189 Refining image 65 img-shot-s00-20130316165138637_0.mtz img-shot-s01-20130316165130255_0.mtz rfactor 0.989836 0.0996581 0.827184 0.312349 0 217 Refining image 81 img-shot-s00-20130316165139178_0.mtz img-shot-s00-20130316165052554_0.mtz rfactor 0.984416 0.133042 0.526552 0.311772 0 222 Refining image 85 img-shot-s01-20130316165043795_1.mtz img-shot-s01-20130316165043795_0.mtz rfactor 0.980657 0.116256 0.734999 0.323394 0 201
Post-refinement parameter table
The parameters used to refine each image are output before each merge. An example is given below. This is also output to a CSV file for each cycle named params_cycle_*.csv
so that the data may be analysed by other programs.
Filename Correl Rsplit Partcorrel Refcount Mosaicity Wavelength Bandwidth hRot kRot rlpSize exp cellA cellB cellC img-shot-s00-20130316165428143_0.mtz 0.996824 0.0729367 0.771081 207 0 1.46288 0.0013 -0.00416609 -0.00638984 3.73011e-05 1.5 105.783 105.783 105.783 img-shot-s01-20130316165101088_0.mtz 0.410825 0.734041 0 179 0 1.46487 0.0013 0.000723228 0.0250253 0.000121815 1.5 105.783 105.783 105.783 img-shot-s00-20130316165238305_0.mtz 0.40738 0.753037 0 183 0 1.45447 0.0013 0.00706399 -0.0082277 0.000150607 1.5 105.783 105.783 105.783 img-shot-s00-20130316165159219_0.mtz 0.990674 0.088927 0.865087 216 0 1.46289 0.0013 -0.00286677 -7.7758e-05 6.17136e-05 1.5 105.783 105.783 105.783 img-shot-s01-20130316165021253_0.mtz 0.991536 0.114612 0.787729 217 0 1.46545 0.0013 -0.0109289 -0.0157625 8.25238e-05 1.5 105.783 105.783 105.783
Next steps
After one is happy with the merging statistics for post-refinement, it is time to make a final merge, with recalculation of sigma values.