Remerging xia2 data to a lower resolution (technical)

A fairly common question I get from people is how to remerge data from xia2 to a lower resolution, without rerunning the whole xia2 process. There are three answers to this. The first is that xia2 does not do this, which is not so helpful. The second is that this will be in the pipeline at some point along with being able to retrospectively change other decisions and have xia2 just run with those. The third, which follows, is that there is a relatively straightforward way of doing this with tools included with xia2 and CCP4.

The Example

The example to be considered here is some myoglobin data, where diffraction went past the edge of the detector, and xia2 gave:


For AUTOMATIC/DEFAULT/SAD
High resolution limit                             1.20    5.39    1.20
Low resolution limit                             39.15   39.15    1.24
Completeness                                     87.8    96.0    18.7
Multiplicity                                      7.4     9.3     1.6
I/sigma                                          43.3   108.4     2.9
Rmerge                                          0.025   0.013   0.206
Rmeas(I)                                        0.031   0.018   0.314
Rmeas(I+/-)                                     0.028   0.014   0.284
Rpim(I)                                         0.011   0.006   0.214
Rpim(I+/-)                                      0.013   0.007   0.195
Wilson B factor                                 6.648
Partial bias                                    0.000   0.000   0.000
Anomalous completeness                           79.4    95.7     8.3
Anomalous multiplicity                            3.6     5.0     1.1
Anomalous correlation                            0.418   0.506  -0.119
Anomalous slope                                 1.345   0.000   0.000
dF/F                                            0.037
dI/s(dI)                                        1.104
Total observations                              424674  6953    1428
Total unique                                    57065   746     879

From processing with xia2 -3d ... Clearly it would be nice to have somewhere close to complete data in the outer shell, so a limit of around 1.35A would be appropriate (looking at the Scala log file.) So, this is what we need to do:

mkdir remerge

cd remerge/

get_ccp4_commands ../LogFiles/AUTOMATIC_DEFAULT_scala.log > runit

The last command reads the scala log file and essentially reconstructs an input file to rerun the job:

HKLIN /tmp/example/DEFAULT/scale/AUTOMATIC_DEFAULT_sorted.mtz

/tmp/example/DEFAULT/scale/AUTOMATIC_DEFAULT_scaled.mtz /tmp/example/DEFAULT/scale/AUTOMATIC_DEFAULT_scaled.mtz

SYMINFO /dls_sw/apps/ccp4/64/6.3.0/13nov2012/ccp4-6.3.0/lib/data/syminfo.lib

bins 20

resolution 1.200000

run 1 batch 1 to 321

resolution run 1 high 1.200000

name run 1 project AUTOMATIC crystal DEFAULT dataset SAD

scales constant

exclude sdmin 2.0

sdcorrection fixsdb noadjust norefine both 1.0 0.0

anomalous on

So we can edit this to do what we want:

scala \

HKLIN /tmp/example/DEFAULT/scale/AUTOMATIC_DEFAULT_sorted.mtz \

HKLOUT remerge.mtz << eof

bins 20

resolution 1.35

run 1 batch 1 to 321

name run 1 project AUTOMATIC crystal DEFAULT dataset SAD

scales constant

exclude sdmin 2.0

sdcorrection fixsdb noadjust norefine both 1.0 0.0

anomalous on

eof

and run:

bash runit

to get

...

                                           Overall  InnerShell  OuterShell

  Low resolution limit                       39.15     39.15      1.39

  High resolution limit                       1.35      6.04      1.35

  Rmerge                                     0.024     0.013     0.108

  Rmerge in top intensity bin                0.014        -         -

  Rmeas (within I+/I-)                       0.027     0.014     0.139

  Rmeas (all I+ & I-)                        0.030     0.018     0.140

  Rpim (within I+/I-)                        0.012     0.007     0.086

  Rpim (all I+ & I-)                         0.010     0.006     0.070

  Fractional partial bias                    0.000     0.000     0.000

  Total number of observations              397951      4729     12624

  Total number unique                        46162       521      3333

  Mean((I)/sd(I))                             52.1     107.9      10.5

  Completeness                                99.7      94.4      97.9

  Multiplicity                                 8.6       9.1       3.8

  Anomalous completeness                      96.6      93.9      78.8

  Anomalous multiplicity                       3.9       4.9       1.9

  DelAnom correlation between half-sets      0.410     0.587     0.115

  Mid-Slope of Anom Normal Probability       1.422       -         -

...

i.e. essentially complete data. To get this output in scalepack format, adjust the script to:

scala \

HKLIN /tmp/example/DEFAULT/scale/AUTOMATIC_DEFAULT_sorted.mtz \

SCALEPACK unmerged.sca \

HKLOUT remerge.mtz << eof

bins 20

output polish unmerged

resolution 1.35

run 1 batch 1 to 321

name run 1 project AUTOMATIC crystal DEFAULT dataset SAD

scales constant

exclude sdmin 2.0

sdcorrection fixsdb noadjust norefine both 1.0 0.0

anomalous on

eof

and run again - now you can see why I usually just advise running xia2 again with the -resolution flag :o)