Hi Georg, 
 
Am Samstag, 09. Januar 2016 10:46 CET, Georg Mlynek <[log in to unmask]> schrieb: 
 
> Dear Kay, could you please explain following points regarding the 
> keyword REFERENCE_DATA_SET= for the CORRECT step
> 
> 1. what is the advantage of using the input keyword REFERENCE_DATA_SET= 
> during processing of a new dataset compared to just scale the datasets 
> later with XSCALE?

it ensures consistent indexing. This is important in many space groups (e.g. P3x P4x P6x) - see column "alternative indexing possible?" in the table at http://strucbio.biologie.uni-konstanz.de/xdswiki/index.php/Space_group_determination .

> 
> 2. If using REFERENCE_DATA_SET= and SPACE_GROUP_NUMBER= are the unit 
> cell dimensions refined during  REFINE(CORRECT)= or are the unit cell 
> dimensions not influenced at all by the REFERENCE_DATA_SET= ?

"yes" is the answer to both questions (these are not alternatives!). The cell _is_ refined, _and_  REFERENCE_DATA_SET does _not_ influence the result of the cell refinement. 

> 
> 3. In the manual one can find "You may specify here the file name of 
> previously measured data from the same crystal form." From your 
> experience does this mean really just the same same crystal form 
> (spacegroup) or should it be crystals from the same drop/condition or 
> datsets from the same crystal. In what cases do you recommend to use 
> REFERENCE_DATA_SET=

I would use it only to ensure consistent indexing if you have one of the spacegroups mentioned above; otherwise you don't  need it (but it would give you inter-dataset R values). It does not matter whether the REFERENCE_DATA_SET is from a crystal of the same drop; it should be your best dataset - the one that you would list first in XSCALE.INP, and which all others would be scaled relative to.
Use of REFERENCE_DATA_SET affects the XDS_ASCII.HKL in three ways: 1) the indexing - this is what you need to get right in case you work in one those spacegroups, 2) an overall scale factor - this is unimportant, 3) an overall (Wilson-type) B factor - this will be applied later in XSCALE anyways. If you don't want to have 2) and 3) applied, then you should run CORRECT twice: first with REFERENCE_DATA_SET, to get the 12 reindexing numbers, secondly without REFERENCE_DATA_SET, but with those 12 values as parameters to REIDX= .

HTH,

Kay 

> 
> Thanks in advance, best regards Georg.
> 
> 
>         REFERENCE_DATA_SET=
> 
> You may specify here the file name of previously measured data from the 
> same crystal form. If available, these data are used by XPLAN and CORRECT.
> XPLAN uses the old data to tell the user by what strategy a maximum of 
> new data could be collected.
> CORRECT uses the old data for local scaling and comparison with the 

> current data set. In case of setting ambiguities the reference data are 
> often used for selecting the best fitting one.
> The reference data set is assumed to possess the symmetry specified by 
> the parameter SPACE_GROUP_NUMBER= in XDS.INP. As a consequence, for an 
> unknown space group, the reference data will be ignored.
> The old, reference data set should be of type XDS_ASCII. If the data 
> cannot be read successfully, XDS assumes that there are no such data 
> available.
> 
> /Example/: REFERENCE_DATA_SET= ../XDS_ASCII_native.HKL
> The file name of a reference data set.
> 
> Parameter is used by XPLAN, CORRECT
> 
> On 12/31/2015 09:03 PM, Kay Diederichs wrote:
> > Dear XDS users,
> >
> > since today, there is a newly built XDS package on the XDS download
> > server which identifies itself as
> > VERSION Oct 15, 2015  BUILT=20151231
> > (this is printed to the terminal window when you run xds or xds_par; it
> > is _not_ in the .LP files)
> > For those of you who want or need to update, this is what you should
> > install.
> >
> > A word of caution: there was (from Nov 19 until today) a version
> > VERSION Oct 15, 2015  BUILT=20151119
> > on the download server which has an issue: it produces data with
> > unrealistically good Rmeas and I/sigI values (as found by two users -
> > thanks for reporting this!). Do not use this version! If you installed
> > BUILT=20151119 you should replace it with the one indicated above, and
> > be very critical about results obtained with BUILT=20151119. Sorry for this!
> >
> > I wish everybody a great 2016!
> >
> > Kay
> >
> 
 
-- 
Kay Diederichs                            http://strucbio.biologie.uni-konstanz.de
email: [log in to unmask] Tel +49 7531 88 4049 Fax 3183
Fachbereich Biologie, Universität Konstanz, Box 647, D-78457 Konstanz