search

helenginn / cppxfel

cppxfel source - I finished my doctorate. This program is FOSSILISED! Assume no more support!!
3 stars 2 forks source link

NANs in anomalous_diff.mtz #10

Open biochem-fan opened 8 years ago

biochem-fan commented 8 years ago

SIGMAs of some reflections in anomalous_diff.mtz are NaN. Thus, iotbx fails to load the file (e.g. phenix.xtriage and phenix.reflection_file_converter).

To use the file in SHELX, the workaround is as follows:

mtz2sca input.mtz
grep -v nan input.sca > fixed.sca
helenginn commented 8 years ago

Workaround for MTZ files:

sftools << EOF read input.mtz select col SIGI = present select col I = present purge yes write input_filtered.mtz exit EOF

helenginn commented 8 years ago

NANs were occurring for at least negative intensities. Have attempted to provide a fix for this and remove Friedel pairs where one is missing observations. Single-observation intensities are currently square-rooted to give sigma values until counting sigmas are carried through the experiment (i.e. read/write data to HDF5).

Please close when you are happy with this fix.

biochem-fan commented 8 years ago

Confirmed. ANODE sigma stayed the same, while the completeness slightly improved. Thank you very much for your fix.

biochem-fan commented 8 years ago

anomalous_diff_N.mtz during REFINE_PARTIALITY is OK but anomalous_diff.mtz after MERGE still contains NANs.

helenginn commented 8 years ago

This does not appear to be a problem anymore with the re-written MtzMerger, which replaces MtzGrouper, accessed by adding OLD_MERGE OFF.