Hi Jacob,
Shouldn't POINTLESS be able to reindex all the datasets to match the indexing of one of them?
As far as I'm aware, if you choose the option "Match index to reference" in POINTLESS and run this on two of your data sets using the third one as the reference set, it will simply reindex the files with a consistent origin and orientation and it will not change the space group you used to integrate the data.
After this you can use the option "Just combine input files" to create a single mtz file that combines all three data sets (your reference file and the two newly reindexed files). You can then scale this file in AIMLESS but remember to select "Customise symmetry determination" and then change it from "Determine Laue group" to "Just sort input files", otherwise AIMLESS will run POINTLESS again and change the space group from P3212 to P622 before scaling the data.
Best,
Tony
------------------------------------------------------
Dr. Antonio Ariza
University of Oxford
Sir William Dunn School of Pathology
South Parks Road
Oxford
OX1 3RE
________________________________________
From: CCP4 bulletin board [[log in to unmask]] on behalf of [log in to unmask] [[log in to unmask]]
Sent: 14 July 2015 16:48
To: [log in to unmask]
Subject: [ccp4bb] AW: Reindexing and Bijvoet Mates
Dear Jacob,
In P3212, there are two equally valid ways to index the data. With the highly twinned data, these two indexing options will be very similar and Mosflm might have choosen different options for the different runs, thereby canceling the anomalous signal. The best way to proceed is to use the orientation matrix of the first run for the other runs as well.
The global phasing people (Clemens Vonrhein) have implemented this in autoPROC, so I really would recommend using this. If you don't like XDS, autoPROC can also use Mosflm.
Best,
Herman
-----Ursprüngliche Nachricht-----
Von: CCP4 bulletin board [mailto:[log in to unmask]] Im Auftrag von Keller, Jacob
Gesendet: Dienstag, 14. Juli 2015 17:17
An: [log in to unmask]
Betreff: [ccp4bb] Reindexing and Bijvoet Mates
Dear Crystallographers,
I am concerned that possibly my data need to be reindexed--here are the facts:
--From refined structure, data is highly twinned ~45%, p3212 spacegroup, taken on home source
--three 180-degree sweeps at 0.1 deg/image were taken, one starting at phi = 0, two taken starting at phi = 180, with one of those having chi = 10 deg.
--when scaled together, there is no appreciable anomalous signal in scaling, even though the data are really high-quality to 1.7 (corner of detector). I realize twinning may be playing a role here, but since data are so good, I would think I would see something. When I use the model against a single dataset, I can see even small anomalous peaks from Cl and S.
If left to its own devices, Pointless puts it in 622, but that's wrong because of the twinning. When told to do p3212, it gives the info below. I am curious--is it possible that the 0 and 180 datasets were indexed such that the Bijvoet differences are cancelling? Is that theoretically possible? Could it happen in mosflm? Autoindexing has always been a black box to me, partially because of the proprietary algorithm in one of the popular indexing programs.
All the best,
Jacob Keller
=======================================================================
Choosing specified space group P 32 1 2 with reindex operator [h,k,l]
Alternative indexing scores relative to reference
Alternative reindexing Lklhd CC R(E^2) Number Cell_deviation
[-h,-k,l] 0.129 0.830 0.164 495755 0.06
[k,h,-l] 0.127 0.828 0.165 445580 0.06
[k,-h-k,l] 0.088 0.791 0.188 428484 0.06
[h,-h-k,-l] 0.084 0.786 0.189 425693 0.06
[h+k,-h,l] 0.084 0.785 0.193 424023 0.08
[-h,h+k,-l] 0.083 0.784 0.191 431379 0.08
[h+k,-k,-l] 0.080 0.779 0.191 435826 0.06
[-h-k,h,l] 0.075 0.773 0.194 434008 0.06
[-h-k,k,-l] 0.070 0.764 0.200 444985 0.08
[-k,h+k,l] 0.068 0.761 0.198 437727 0.08
[h,k,l] 0.057 0.737 0.212 454077 0.08
[-k,-h,-l] 0.056 0.735 0.210 466665 0.07
Alternative indexing scores relative to reference
Alternative reindexing Lklhd CC R(E^2) Number Cell_deviation
[-h,-k,l] 0.166 0.890 0.126 568040 0.07
[k,h,-l] 0.100 0.848 0.135 525759 0.07
[h,-h-k,-l] 0.092 0.839 0.155 513814 0.07
[-h,h+k,-l] 0.091 0.839 0.156 510382 0.08
[h+k,-k,-l] 0.077 0.823 0.159 527763 0.06
[h+k,-h,l] 0.074 0.819 0.158 508182 0.08
[-h-k,k,-l] 0.071 0.815 0.161 527394 0.08
[-k,h+k,l] 0.071 0.815 0.162 519381 0.07
[-h-k,h,l] 0.071 0.814 0.158 507974 0.07
[h,k,l] 0.069 0.812 0.169 540345 0.08
[k,-h-k,l] 0.063 0.802 0.163 516689 0.06
[-k,-h,-l] 0.056 0.789 0.171 538172 0.07
Alternative indexing relative to first file(s):
Reindex operator CC Lklhd Confidence File name
2 [-h,-k,l] 0.830 0.129 0.016 DataSetAinv_p1.mtz
3 [-h,-k,l] 0.890 0.166 0.104 DataSetAchi10_p1.mtz
Chosen Solution: space group P 32 1 2
Reindex operator: [h,k,l]
Laue group probability: 0.004
Systematic absence probability: 0.996
Total probability: 0.004
Space group confidence: 1.000
Laue group confidence 1.000
Unit cell: 98.89 98.91 128.55 90.00 89.98 119.98
23.08 to 1.73 - Resolution range used for Laue group search
23.08 to 1.70 - Resolution range in file, used for systematic absence check
*******************************************
Jacob Pearson Keller, PhD
Looger Lab/HHMI Janelia Research Campus
19700 Helix Dr, Ashburn, VA 20147
email: [log in to unmask]
*******************************************
