WORK-IN-PROGRESS
The data was collected on 18th March 2020 at beamline 19-ID (SBC-CAT, APS).
We have two distinct sweeps of data available:
y3_b07_#####.cbf : 1 - 60
y3_b07b_#####.cbf : 2 - 140
At first sight this seems a slightly unusual strategy given the characteristics of the detector used (Förster et al, 2019). We would usually want to see low-dose, high-multiplicity, fine-sliced data collected at modern synchrotrons with the current generation of detectors (Pflugrath, 1999, Mueller at al, 2012).
It is also unfortunate that the second sweep was started again at the same Omega angle (–90) as the first one. So although 100 degrees of data was collected, only 70 degrees of rotation is covered - which for monoclinic C2 results in very poor overall completeness (but then: even 100 degrees would have been problematic): see e.g. PDBpeep.
The images have a very high average background value of about 40 counts
This seems to suggest that there could have been at least a factor of 10 reduction in dose per 0.5 degree image. A better sampling of the reflection profile (through fine-slicing) would have been possible by using 0.1 degree per image with an exposure time of 0.1 seconds: this would have put the same amount of dose into the crystal as the current 0.5sec/0.5deg. The Pilatus3 6M can run at a frame rate of 25Hz, i.e. at 0.04 sec/image). So one could collect data as
sweep 1: 0.1 deg/image
0.05 sec/image
100% transmission
60 degrees of data
start at Omega=-90
sweep 2: 0.1 deg/image
0.05 sec/image
100% transmission
140 degrees of data
start at Omega=-30
which would have put the exact same dose into the crystal with the same collection wall clock time - but would result in a low-dose and fine-sliced dataset of adequate multiplicity (200 degrees of rotation covered instead of 70 degrees). It is most likely, that halving the transmission would have been a better strategy on top of this (achieving high multiplicity via 400 degrees of data on top of low-dose and fine-slicing), so
sweep 1: 0.1 deg/image
0.05 sec/image
50% transmission
120 degrees of data
start at Omega=-90
sweep 2: 0.1 deg/image
0.05 sec/image
50% transmission
280 degrees of data
start at Omega=+30
The total collection time would have been longer though: 200 seconds instead of 100 seconds. but given the fact that it took 224 seconds to recenter (?) the crystal between sweep 1 and 2, those additional 100 seconds of time do not seem crucial. If we assume that the initial centering took a similar amount of time, we have for the current data collection
sweep 1 centering : 225 sec
collection : 30 sec
sweep 2 centering : 225 sec
collection : 70 sec
------------------------------
total 550 sec = 9 min 10 sec
and a slightly different strategy described above:
sweep 1 centering : 225 sec
collection : 60 sec
sweep 2 centering : 225 sec
collection : 140 sec
------------------------------
total 650 sec = 10 min 50 sec
Remember: we can always glue fine-sliced images together (creating 0.5 degree images out of five 0.1 degree images) - but we can never split images collected in wide-slicing mode.
After downloading the images (and placing the mini-cbf files into a directory Images), we can run autoPROC on that dataset via
process ReverseRotationAxis=yes -I Images -d 00 | tee 00.lisThe extra keyword is recorded on our beamline settings wiki page.
Both sweeps of data show strong indications of radiation damage:
| Sweep | Spots found | R-values |
| 1 |  |
 |
| 2 |  |
 |
If these had been low-dose, high-multiplicity datasets we could have (1) restricted the image ranges to the earlier half for each sweep, or (2) analysed the data for early and late parts (and use those results for F(early)–F(late) radiation-damage detection maps in BUSTER).
The problem with missing data is visible in the STARANISO plots:
 |
 |
 |
and the merging statistics:
anisotropic (STARANISO) analysis:
Spacegroup name C2
Unit cell parameters 190.8751 110.2789 64.2203 90.0000 96.2532 90.0000
Wavelength 0.97918 A
Diffraction limits & principal axes of ellipsoid fitted to diffraction cut-off surface:
2.503 0.7227 0.0000 0.6912 0.962 _a_* + 0.273 _c_*
2.168 0.0000 1.0000 0.0000 _b_*
3.488 -0.6912 0.0000 0.7227 -0.933 _a_* + 0.360 _c_*
Number of active ice-rings within this resolution range = 0
Number of RUNs (sweeps) contributing to this dataset = 2
Criteria used in determination of diffraction limits:
-----------------------------------------------------
local(I/sigI) >= 1.20
Overall InnerShell OuterShell
---------------------------------------------------------------------------
Low resolution limit 63.838 63.838 2.571
High resolution limit 2.207 6.865 2.207
Rmerge (all I+ & I-) 0.161 0.060 0.637
Rmerge (within I+/I-) 0.141 0.058 0.582
Rmeas (all I+ & I-) 0.197 0.073 0.845
Rmeas (within I+/I-) 0.190 0.078 0.802
Rpim (all I+ & I-) 0.110 0.040 0.547
Rpim (within I+/I-) 0.126 0.053 0.548
Total number of observations 56301 3992 1823
Total number unique 20852 1408 849
Mean(I)/sd(I) 8.8 26.3 1.5
Completeness (spherical) 31.4 61.6 3.5
Completeness (ellipsoidal) 47.2 61.6 8.4
Multiplicity 2.7 2.8 2.1
CC(1/2) 0.983 0.994 0.428
traditional (isotropic) analysis:
Overall InnerShell OuterShell
---------------------------------------------------------------------------
Low resolution limit 51.754 51.754 2.697
High resolution limit 2.652 7.192 2.652
Rmerge (all I+ & I-) 0.205 0.057 1.264
Rmeas (all I+ & I-) 0.251 0.070 1.785
Rpim (all I+ & I-) 0.142 0.040 1.261
Total number of observations 64946 3441 1130
Total number unique 25685 1226 961
Mean(I)/sd(I) 6.7 29.5 0.9
Completeness 66.8 62.0 50.0
Multiplicity 2.5 2.8 1.2
CC(1/2) 0.975 0.994 0.370
For comparison the values from the deposited structure (as available):
Overall InnerShell OuterShell
---------------------------------------------------------------------------
Low resolution limit 43.680 43.680 2.750
High resolution limit 2.700 7.320 2.700
Rmerge (all I+ & I-) 0.140 0.060 0.637
Rmeas (all I+ & I-) 0.173 0.071 0.769
Rpim (all I+ & I-) 0.099 0.040 0.492
Total number unique 20799 1163 679
Mean(I)/sd(I) 4.9 NA NA
Completeness 57.3 62.6 38.4
Multiplicity 2.5 2.9 1.9
CC(1/2) NA 0.991 0.566
These show better R-values (but poorer <I/sigma(I)> values) and lower completeness: so maybe some images (towards the end of each sweep where radiation damage becomes very pronounced) were exclued? This would also explain the better R-values (but not the lower <I/sigma(I)> value).