Click here to go to the corresponding page for the latest version of DIALS
Multi-lattice Tutorial¶
Introduction¶
The following example uses semi-synthetic multi-lattice trypsin datasets
collected using beamline I04 at Diamond Light Source which is available for
download from 
. This tutorial uses one of the 2-lattice
datasets which are contained in the file semisynthetic_multilattice_data_2.tar.bz2.
In this tutorial we shall focus on the processing steps that diverge from the regular single-lattice processing discussed in Processing in Detail.
Import and Spotfinding¶
As for single-lattice processing, the first steps are to import the data and find spots using the following commands:
dials.import semisynthetic_multilattice_data/2/ag/trp_ag_*.cbf
dials.find_spots imported.expt min_spot_size=3 nproc=4
During import, all that happens here is that the image headers are read, and a
file describing their contents (imported.expt) is written. The output
just describes what the software understands of the images it was
passed, in this case one sequence of data containing 100 images:
The following parameters have been modified:
input {
experiments = <image files>
}
--------------------------------------------------------------------------------
format: <class 'dxtbx.format.FormatCBFMiniPilatusDLS6MSN100.FormatCBFMiniPilatusDLS6MSN100'>
num images: 100
sequences:
still: 0
sweep: 1
num stills: 0
--------------------------------------------------------------------------------
Writing experiments to imported.expt
For the spot finding, we tweak the minimum spot size (min_spot_size=3) to improve the results for this dataset and use multiple processors to speed up the spot-finding (nproc=4):
Extracted 46332 spots
Removed 8863 spots with size < 3 pixels
Removed 3 spots with size > 1000 pixels
Calculated 37466 spot centroids
Calculated 37466 spot intensities
Filtered 35422 of 37466 spots by peak-centroid distance
Histogram of per-image spot count for imageset 0:
35422 spots found on 100 images (max 1190 / bin)
*
* *
* *
* *
** *
**** ******************* * ************* ******** ** *******
************************************************************
************************************************************
************************************************************
************************************************************
1 image 100
--------------------------------------------------------------------------------
Saved 35422 reflections to strong.refl
Indexing¶
The next step is the indexing of the strong spots. By default only one lattice is searched for, but if there are sufficient unindexed reflections remaining after indexing the first lattice, we can switch on indexing of multiple lattices using the parameter max_lattices=2 (e.g.):
dials.index imported.expt strong.refl max_lattices=2
RMSDs by experiment:
---------------------------------------------
| Exp | Nref | RMSD_X | RMSD_Y | RMSD_Z |
| id | | (px) | (px) | (images) |
---------------------------------------------
| 0 | 1000 | 0.41832 | 0.25232 | 0.15582 |
| 1 | 1000 | 0.33596 | 0.24331 | 0.17531 |
---------------------------------------------
Refined crystal models:
model 1 (16636 reflections):
Crystal:
Unit cell: (54.063(4), 58.2475(18), 66.494(2), 89.9778(13), 90.013(3), 90.012(3))
Space group: P 1
U matrix: {{ 0.1870, 0.7632, -0.6185},
{ 0.0427, 0.6227, 0.7813},
{ 0.9814, -0.1725, 0.0838}}
B matrix: {{ 0.0185, 0.0000, 0.0000},
{ 0.0000, 0.0172, 0.0000},
{ 0.0000, -0.0000, 0.0150}}
A = UB: {{ 0.0035, 0.0131, -0.0093},
{ 0.0008, 0.0107, 0.0117},
{ 0.0182, -0.0030, 0.0013}}
model 2 (17247 reflections):
Crystal:
Unit cell: (54.080(3), 58.263(2), 66.498(3), 90.0060(18), 90.021(3), 90.045(3))
Space group: P 1
U matrix: {{ 0.0094, 0.6714, -0.7410},
{ 0.3813, -0.6875, -0.6180},
{-0.9244, -0.2768, -0.2625}}
B matrix: {{ 0.0185, 0.0000, 0.0000},
{ 0.0000, 0.0172, 0.0000},
{ 0.0000, 0.0000, 0.0150}}
A = UB: {{ 0.0002, 0.0115, -0.0111},
{ 0.0070, -0.0118, -0.0093},
{-0.0171, -0.0048, -0.0039}}
--------------------------------------------------
| Imageset | # indexed | # unindexed | % indexed |
--------------------------------------------------
| 0 | 33883 | 1539 | 95.7% |
--------------------------------------------------
Change of basis op: a,b,c
Rotation matrix to transform crystal 1 to crystal 2:
{{0.973, -0.160, -0.169},
{-0.071, -0.895, 0.441},
{-0.222, -0.417, -0.881}}
Rotation of -154.399 degrees about axis (0.993, -0.061, -0.103)
Saving refined experiments to indexed.expt
Saving refined reflections to indexed.refl
Next we run dials.refine_bravais_settings refining each indexing solution (separately) in all Bravais settings consistent with the indexed unit cell. In this example we would continue processing using bravais_setting_5.expt, i.e. solution number 5.
dials.refine_bravais_settings indexed.expt indexed.refl crystal_id=0
dials.refine_bravais_settings indexed.expt indexed.refl crystal_id=1
gives a table containing the metric fit, rmsds (in mm) and unit cell for each Bravais setting…
----------------------------------------------------------------------------------------------------------------
Solution Metric fit rmsd min/max cc #spots lattice unit_cell volume cb_op
----------------------------------------------------------------------------------------------------------------
9 4.2490 1.579 0.384/0.763 1000 tP 60.31 60.31 69.15 90.00 90.00 90.00 251517 a,b,c
8 4.2490 1.508 0.372/0.529 1000 oC 85.65 84.55 69.06 90.00 90.00 90.00 500054 a+b,-a+b,c
7 4.2490 1.487 0.372/0.372 1000 mC 84.57 85.57 69.01 90.00 89.92 90.00 499425 a-b,a+b,c
6 4.2490 1.560 0.529/0.529 1000 mC 85.71 84.31 69.03 90.00 89.87 90.00 498876 a+b,-a+b,c
* 5 0.0000 0.095 0.245/0.904 1000 oP 54.10 58.27 66.51 90.00 90.00 90.00 209657 a,b,c
* 4 0.0000 0.088 0.904/0.904 1000 mP 58.27 54.11 66.52 90.00 89.98 90.00 209735 -b,-a,-c
* 3 0.0000 0.095 0.245/0.245 1000 mP 54.11 58.27 66.51 90.00 90.02 90.00 209715 a,b,c
* 2 0.0000 0.093 0.384/0.384 1000 mP 54.11 66.52 58.28 90.00 90.02 90.00 209787 -a,-c,-b
* 1 0.0000 0.090 -/- 1000 aP 54.11 58.27 66.51 89.98 90.02 90.00 209725 a,b,c
----------------------------------------------------------------------------------------------------------------
----------------------------------------------------------------------------------------------------------------
Solution Metric fit rmsd min/max cc #spots lattice unit_cell volume cb_op
----------------------------------------------------------------------------------------------------------------
9 4.2639 1.740 0.227/0.833 1000 tP 59.33 59.33 68.32 90.00 90.00 90.00 240463 a,b,c
8 4.2639 1.709 0.242/0.833 1000 oC 84.84 83.85 68.54 90.00 90.00 90.00 487536 a+b,-a+b,c
7 4.2639 1.343 0.631/0.631 1000 mC 83.78 82.53 68.28 90.00 88.90 90.00 472000 a-b,a+b,c
6 4.2639 1.569 0.242/0.242 1000 mC 82.84 83.47 67.49 90.00 91.40 90.00 466492 a+b,-a+b,c
* 5 0.0497 0.076 0.658/0.833 1000 oP 54.10 58.29 66.52 90.00 90.00 90.00 209775 a,b,c
* 4 0.0497 0.078 0.658/0.658 1000 mP 58.29 54.10 66.52 90.00 89.99 90.00 209773 -b,-a,-c
* 3 0.0453 0.075 0.811/0.811 1000 mP 54.09 58.29 66.51 90.00 90.02 90.00 209673 a,b,c
* 2 0.0221 0.075 0.833/0.833 1000 mP 54.09 66.51 58.27 90.00 90.03 90.00 209642 -a,-c,-b
* 1 0.0000 0.075 -/- 1000 aP 54.08 58.27 66.50 90.01 90.02 90.03 209577 a,b,c
----------------------------------------------------------------------------------------------------------------
Now we re-run the indexing, this time imposing the lattice constraints for the chosen Bravais setting, in this case number 5, i.e. oP, or point group P222.
dials.index imported.expt strong.refl max_lattices=2 space_group=P222
RMSDs by experiment:
---------------------------------------------
| Exp | Nref | RMSD_X | RMSD_Y | RMSD_Z |
| id | | (px) | (px) | (images) |
---------------------------------------------
| 0 | 1000 | 0.45694 | 0.26413 | 0.16573 |
| 1 | 1000 | 0.35593 | 0.27804 | 0.20011 |
---------------------------------------------
Refined crystal models:
model 1 (16635 reflections):
Crystal:
Unit cell: (54.100(4), 58.2684(17), 66.517(2), 90.0, 90.0, 90.0)
Space group: P 2 2 2
U matrix: {{ 0.1873, 0.7630, -0.6186},
{ 0.0429, 0.6228, 0.7812},
{ 0.9814, -0.1728, 0.0839}}
B matrix: {{ 0.0185, 0.0000, 0.0000},
{-0.0000, 0.0172, 0.0000},
{-0.0000, 0.0000, 0.0150}}
A = UB: {{ 0.0035, 0.0131, -0.0093},
{ 0.0008, 0.0107, 0.0117},
{ 0.0181, -0.0030, 0.0013}}
model 2 (17249 reflections):
Crystal:
Unit cell: (54.117(3), 58.2882(15), 66.526(2), 90.0, 90.0, 90.0)
Space group: P 2 2 2
U matrix: {{ 0.0091, 0.6714, -0.7410},
{ 0.3810, -0.6874, -0.6182},
{-0.9245, -0.2768, -0.2621}}
B matrix: {{ 0.0185, 0.0000, 0.0000},
{-0.0000, 0.0172, 0.0000},
{ 0.0000, 0.0000, 0.0150}}
A = UB: {{ 0.0002, 0.0115, -0.0111},
{ 0.0070, -0.0118, -0.0093},
{-0.0171, -0.0047, -0.0039}}
--------------------------------------------------
| Imageset | # indexed | # unindexed | % indexed |
--------------------------------------------------
| 0 | 33884 | 1538 | 95.7% |
--------------------------------------------------
Change of basis op: -a,b,-c
Rotation matrix to transform crystal 1 to crystal 2:
{{0.052, 0.997, -0.063},
{-0.978, 0.038, -0.203},
{-0.200, 0.072, 0.977}}
Rotation of -88.056 degrees about axis (-0.138, -0.069, 0.988)
Saving refined experiments to indexed.expt
Saving refined reflections to indexed.refl
Refinement and Integration¶
After indexing, processing proceeds similarly to the single-lattice case. First, the crystal models can be further refined with a scan varying model, in this example also using the tukey outlier rejection algorithm:
dials.refine indexed.expt indexed.refl outlier.algorithm=tukey
Refinement steps:
------------------------------------------------
| Step | Nref | RMSD_X | RMSD_Y | RMSD_Phi |
| | | (mm) | (mm) | (deg) |
------------------------------------------------
| 0 | 2000 | 0.079758 | 0.046104 | 0.018187 |
| 1 | 2000 | 0.066176 | 0.042452 | 0.017411 |
| 2 | 2000 | 0.065727 | 0.042236 | 0.016897 |
| 3 | 2000 | 0.065412 | 0.042413 | 0.016653 |
| 4 | 2000 | 0.065282 | 0.042591 | 0.016592 |
| 5 | 2000 | 0.065257 | 0.042631 | 0.016585 |
| 6 | 2000 | 0.065253 | 0.042632 | 0.016585 |
------------------------------------------------
RMSD no longer decreasing
RMSDs by experiment:
----------------------------------------------
| Exp | Nref | RMSD_X | RMSD_Y | RMSD_Z |
| id | | (px) | (px) | (images) |
----------------------------------------------
| 0 | 13833 | 0.48806 | 0.27504 | 0.1626 |
| 1 | 14752 | 0.35492 | 0.2951 | 0.20587 |
----------------------------------------------
Updating predictions for indexed reflections
Saving refined experiments to refined.expt
Saving reflections with updated predictions to refined.refl
Next, we integrate the data:
dials.integrate refined.expt refined.refl
This program outputs a lot of information as integration progresses, concluding with a summary of the integration results for each lattice:
Summary for experiment 0
---------------------------------------------------------------
Item | Overall | Low | High
---------------------------------------------------------------
dmin | 1.06 | 2.87 | 1.06
dmax | 43.87 | 43.87 | 1.08
number fully recorded | 25027 | 1859 | 29
number partially recorded | 9850 | 760 | 6
number with invalid background pixels | 10448 | 623 | 29
number with invalid foreground pixels | 5595 | 394 | 17
number with overloaded pixels | 4 | 4 | 0
number in powder rings | 0 | 0 | 0
number processed with summation | 29114 | 2208 | 18
number processed with profile fitting | 23507 | 1796 | 6
number failed in background modelling | 20 | 5 | 0
number failed in summation | 5595 | 394 | 17
number failed in profile fitting | 11202 | 806 | 29
ibg | 17.81 | 49.24 | 4.31
i/sigi (summation) | 26.91 | 151.98 | 1.32
i/sigi (profile fitting) | 35.42 | 199.63 | 1.28
cc prf | 0.94 | 0.85 | 0.98
cc_pearson sum/prf | 0.89 | 0.86 | 0.93
cc_spearman sum/prf | 0.97 | 0.98 | 0.37
---------------------------------------------------------------
Summary for experiment 1
---------------------------------------------------------------
Item | Overall | Low | High
---------------------------------------------------------------
dmin | 1.06 | 2.87 | 1.06
dmax | 25.49 | 25.49 | 1.08
number fully recorded | 24816 | 1800 | 38
number partially recorded | 10172 | 818 | 9
number with invalid background pixels | 9013 | 539 | 34
number with invalid foreground pixels | 5147 | 369 | 23
number with overloaded pixels | 3 | 3 | 0
number in powder rings | 0 | 0 | 0
number processed with summation | 29676 | 2237 | 24
number processed with profile fitting | 24069 | 1812 | 11
number failed in background modelling | 85 | 29 | 0
number failed in summation | 5147 | 369 | 23
number failed in profile fitting | 10754 | 794 | 36
ibg | 17.79 | 49.75 | 4.36
i/sigi (summation) | 24.71 | 139.98 | 1.44
i/sigi (profile fitting) | 32.19 | 182.76 | 1.77
cc prf | 0.94 | 0.84 | 0.96
cc_pearson sum/prf | 0.67 | 0.60 | 0.95
cc_spearman sum/prf | 0.97 | 0.97 | 0.70
---------------------------------------------------------------
Symmetry, Scaling and Merging¶
Again, we can proceed as standard, with the programs handling the multiple lattices found in the datafiles:
dials.symmetry integrated.expt integrated.refl
Scoring all possible sub-groups
---------------------------------------------------------------------------------------------
Patterson group Likelihood NetZcc Zcc+ Zcc- CC CC- delta Reindex operator
---------------------------------------------------------------------------------------------
P m m m *** 0.988 9.87 9.87 0.00 0.99 0.00 0.0 a,b,c
P 1 2/m 1 0.004 0.13 9.93 9.80 1.00 0.98 0.0 -b,-a,-c
P 1 2/m 1 0.004 0.10 9.92 9.81 1.00 0.98 0.0 -a,-c,-b
P 1 2/m 1 0.004 0.03 9.88 9.85 1.00 0.99 0.0 a,b,c
P -1 0.000 0.17 9.99 9.82 1.00 0.98 0.0 a,b,c
---------------------------------------------------------------------------------------------
Best solution: P m m m
Unit cell: (54.1104, 58.2822, 66.5198, 90, 90, 90)
Reindex operator: a,b,c
Laue group probability: 0.988
Laue group confidence: 0.986
...
Laue group: P m m m
---------------------------------------------------------------------------------------------------------------
| Screw axis | Score | No. present | No. absent | <I> present | <I> absent | <I/sig> present | <I/sig> absent |
---------------------------------------------------------------------------------------------------------------
| 21a | 0.000 | 0 | 0 | 0.000 | 0.000 | 0.000 | 0.000 |
| 21b | 1.000 | 9 | 10 | 16592.455 | 13.316 | 141.588 | 0.412 |
| 21c | 1.000 | 12 | 13 | 28257.845 | 2.137 | 150.792 | 0.055 |
---------------------------------------------------------------------------------------------------------------
------------------------
| Space group | score |
------------------------
| P 2 2 2 | 0.0000 |
| P 2 2 21 | 0.0000 |
| P 2 21 2 | 0.0000 |
| P 21 2 2 | 0.0000 |
| P 21 21 2 | 0.0000 |
| P 21 2 21 | 0.0000 |
| P 2 21 21 | 1.0000 |
| P 21 21 21 | 0.0000 |
------------------------
Recommended space group: P 2 21 21
The symmetry analysis suggested space group P 2 21 21, however it is worth noting that no reflections were available to test the 21a screw axis, so this possibility should also be tested during structure solution.
Next we scale the data and inspect the results from the log output or the
scaling.html generated html report:
dials.scale symmetrized.expt symmetrized.refl
----------Merging statistics----------
Resolution: 28.89 - 1.06
Observations: 46152
Unique reflections: 34551
Redundancy: 1.3
Completeness: 36.52%
Mean intensity: 2288.5
Mean I/sigma(I): 15.4
R-merge: 0.037
R-meas: 0.051
R-pim: 0.035
Statistics by resolution bin:
d_max d_min #obs #uniq mult. %comp <I> <I/sI> r_mrg r_meas r_pim cc1/2 cc_ano
28.90 2.89 3375 2366 1.43 46.98 15165.9 43.1 0.030 0.042 0.028 0.993* 0.725
2.89 2.29 3400 2285 1.49 47.13 5801.2 36.6 0.031 0.042 0.028 0.992* -0.239
2.29 2.00 3369 2270 1.48 47.57 4059.8 31.6 0.035 0.046 0.030 0.993* -0.235
2.00 1.82 3297 2143 1.54 44.86 2512.9 26.2 0.039 0.053 0.036 0.986* -0.064
1.82 1.69 3226 2255 1.43 47.61 1506.1 19.6 0.049 0.066 0.045 0.983* 1.000
1.69 1.59 3246 2372 1.37 50.22 1133.3 15.7 0.059 0.081 0.055 0.978* 0.000
1.59 1.51 3229 2387 1.35 50.31 840.0 12.4 0.072 0.100 0.068 0.969* -0.345
1.51 1.45 3289 2481 1.33 52.63 659.5 10.0 0.081 0.112 0.077 0.962* -0.561
1.45 1.39 3177 2494 1.27 53.10 499.8 7.7 0.082 0.114 0.079 0.959* -1.000
1.39 1.34 3082 2480 1.24 52.71 409.9 6.3 0.095 0.132 0.091 0.937* 0.000
1.34 1.30 3207 2540 1.26 54.03 371.5 5.6 0.089 0.124 0.085 0.973* 0.000
1.30 1.26 2878 2306 1.25 49.25 331.8 4.9 0.097 0.136 0.095 0.956* 0.000
1.26 1.23 2272 1845 1.23 39.23 313.8 4.6 0.077 0.108 0.076 0.977* 0.000
1.23 1.20 1632 1293 1.26 27.46 286.4 4.1 0.075 0.105 0.074 0.979* 0.000
1.20 1.17 1217 997 1.22 21.36 267.1 3.8 0.072 0.102 0.072 0.982* 0.000
1.17 1.15 893 775 1.15 16.66 235.7 3.3 0.069 0.097 0.069 0.981* 0.000
1.15 1.12 651 589 1.11 12.60 211.3 2.8 0.074 0.104 0.074 0.981* 0.000
1.12 1.10 425 400 1.06 8.55 166.5 2.2 0.134 0.189 0.134 0.943* 0.000
1.10 1.08 237 225 1.05 4.83 171.0 2.2 0.125 0.176 0.125 0.912* 0.000
1.08 1.06 50 48 1.04 1.02 151.9 1.9 0.197 0.278 0.197 1.000 0.000
28.89 1.06 46152 34551 1.34 36.52 2288.5 15.4 0.037 0.051 0.035 0.995* 0.445*
If required, we can rerun scaling with a resolution limit using the option
d_min=, however in this case the CC1/2 and <I/sI> are reasonable
to the highest resolution measured. The “Analysis by image number” plots
in the scaling.html report also indicate that both datasets are of similar
quality.
Once we are happy with the scaled dataset, a merged MTZ file can be generated:
dials.merge scaled.expt scaled.refl
Writing reflections to merged.mtz
Title: From dials.merge
Space group symbol from file: P22121
Space group number from file: 18
Space group from matrices: P 2 21 21 (No. 18)
Point group symbol from file: 222
Number of crystals: 1
Number of Miller indices: 32658
Resolution range: 28.8871 1.06499
History:
From DIALS 2.dev.1041-gf88516da7, run on 2019-10-28 at 15:40:44 GMT
Crystal 1:
Name: XTAL
Project: AUTOMATIC
Id: 1
Unit cell: (54.1104, 58.2822, 66.5198, 90, 90, 90)
Number of datasets: 1
Dataset 1:
Name: NATIVE
Id: 1
Wavelength: 0.97949
Number of columns: 17
label #valid %valid min max type
H 32658 100.00% 0.00 37.00 H: index h,k,l
K 32658 100.00% 0.00 53.00 H: index h,k,l
L 32658 100.00% 0.00 60.00 H: index h,k,l
IMEAN 32658 100.00% -109.31 213880.50 J: intensity
SIGIMEAN 32658 100.00% 5.93 5046.71 Q: standard deviation
I(+) 24848 76.09% -109.31 213880.50 K: I(+) or I(-)
SIGI(+) 24848 76.09% 5.93 5046.71 M: standard deviation
I(-) 12695 38.87% -93.28 180658.78 K: I(+) or I(-)
SIGI(-) 12695 38.87% 11.23 4263.08 M: standard deviation
N(+) 24848 76.09% 2.00 4.00 I: integer
N(-) 12695 38.87% 4.00 4.00 I: integer
F 32658 100.00% 2.08 460.34 F: amplitude
SIGF 32658 100.00% 0.25 27.28 Q: standard deviation
F(+) 24848 76.09% 2.08 460.34 G: F(+) or F(-)
SIGF(+) 24848 76.09% 0.25 5.48 L: standard deviation
F(-) 12695 38.87% 2.51 421.81 G: F(+) or F(-)
SIGF(-) 12695 38.87% 0.26 5.05 L: standard deviation
This program also performs a truncation, giving a set of merged intensities and
strictly-positive structure factors (Fs) suitable for downstream structure solution.
In the mtz file, both lattices are combined to give a single dataset. To generate
individual mtz files, one could first use dials.split_experiments before
merging with dials.merge:
dials.split_experiments scaled.expt scaled.refl
input {
experiments = scaled.expt
reflections = scaled.refl
}
Saving experiment 0 to split_0.expt
Saving reflections for experiment 0 to split_0.refl
Saving experiment 1 to split_1.expt
Saving reflections for experiment 1 to split_1.refl
It is worth noting that the splitting of experiments can be performed on the multi-lattice datafiles at any point during the processing if desired.
