[phenixbb] iMosflm .mtz

Nathaniel Echols NEchols at lbl.gov
Thu Feb 11 10:28:58 PST 2010

On Feb 11, 2010, at 5:55 AM, anon wrote:
> i was under the impression Phenix (AutoSol) could handle .mtz files from iMosflm, however when i input the .mtz file into AutoSol, no labels appears in the pull-down menu, and pressing Run results in a dialogue box appearing stating you must select a column label for the reflection file, if no labels are present it may not be a valid format. 

Graeme's answer is absolutely correct - Phenix doesn't handle unscaled data, and basically every program assumes that it has already been scaled and merged. (The partial exception: Scalepack files created with "no merge original index" can be used, but as far as I'm aware every program but SOLVE will merge the data internally.  I'm not sure whether MOSFLM/SCALA/etc. can produce similar data, but I assume the same rules apply)

> If i input the iMosflm .mtz file into phenix.reflection_file_editor, and copy over the I, SIGI, M_ISYM columns, i get the following error; 
>   File "/usr/local/phenix-1.6-289/cctbx_project/iotbx/reflection_file_editor.py", line 358, in __init__
>     "related reflections are not identical.)" % array_name)
> TypeError: not all arguments converted during string formatting

To begin with, anything coming out of MOSFLM without scaling is going to be mangled - it isn't designed to deal with that kind of data.  However, there is a secondary problem, which is that some columns that are created during data processing still cause problems even if you have scaled and merged data.  Specifically, columns like M/ISYM are integers, and may differ between symmetry-related reflections.  This causes problems during some of the internal transformations performed by the reflection file editor.  The error message you should have seen was this:

"The data in <array_name> cannot be merged because it is in integer format.  If you wish to change symmetry (or the input data is unmerged), you must omit this array.  (Note also that merging will fail for R-free flags if the flags for symmetry-related reflections are not identical.)"

[ Unfortunately, I left out some parentheses, so you got the bug message instead. :( ]

> Also, how do you delete individual Jobs in the Job History window (ie jobs that have failed or you no longer want to keep  etc)

You can't (this is deliberate - the complete job history information is used internally).  You can flag individual jobs that you *do* want to see, and tell it to show only those jobs, or you can tell it not to display failed or aborted jobs - both of these mechanisms are intended as a substitute for deleting jobs.


Nathaniel Echols
Lawrence Berkeley Lab
NEchols at lbl.gov

More information about the phenixbb mailing list