I hadn't realized phenix.refine handles this differently.
If I start with an mtz file from truncate, with F, SIGF, IMEAN, SIGIMEAN all available,
the default for phenix.refine seems to be to work with IMEAN, SIGIMEAN.
From the log file . . .
Intensities converted to amplitudes for use in refinement.
Number of F-obs in resolution range: 266875
Number of F-obs <= 0: 22532
Refinement resolution range: d_max = 25.4557
d_min = 2.0000
Fobs statistics after all cutoffs applied:
Miller array info: None
Observation type: xray.amplitude
Type of data: double, size=244343
And then later in the log . . .
============================== Outliers rejection =============================
basic_wilson_outliers = 186
extreme_wilson_outliers = 128
beamstop_shadow_outliers = 9
total = 197
And this gives 244146 reflections to use in refinement.
So it looks as if the weak data simply disappears.
Pete
