Pointless doesn't do this correctly at present - I need to fix it
For those crystals you are merging into a single dataset, you do need
to decide somehow on a single average unit cell. Pointless attempts to
do this by replacing the cell by the average over all batches within
that dataset, but at present I think it gets it wrong with multiple
datasets.
I'll look into it
Phil
On 29 Jul 2008, at 23:07, hari jayaram wrote:
> Hello Phil Evans,
> Thanks a lot for your reply. Pointless worked great to edit XNAME
> and DNAME for unmerged mtz files.
>
> We are now taking multiple crystal data as unmerged mosflm output
> mtzs , squishing them together into one pseudo-crystal mtz with the
> same XNAME and DNAME using Pointless. The mtz header does have
> correctly 1 dataset. The only issue I guess is the CELL parameters
> are from one of the two datasets.
>
> This does seem to be quite strange but we are trying to mimic the
> "Scenario 4" as presented in the HKL-scalepack manual in the section
> combining multiple native datasets together . In that scenario it
> asks you to take multiple native data-sets together and combine
> their *.x files and fit one a*, b* and c* and a batch mosaicity ,
> rotx , roty .
>
>
> Wondering what is the correct way of getting the CELL parameters
> for this multi-crystal dataset.
>
>
> Hari
>
>
>
>
> On Tue, Jul 29, 2008 at 5:06 PM, Phil Evans <[log in to unmask]>
> wrote:
> The easiest way to combine multiple crystals and renaming datasets
> for Scala is to use Pointless, available from the CCP4 prerelease
> site. You can't use CAD on unmerged files: the older programs
> Rebatch & Sortmtz can also be used
>
> Note the if you are merging multiple crystals, you are essentially
> creating a composite "crystal" so all parts to be merged must have
> the same XNAME & DNAME
>
> Phil
>
>
>
>
>
> On 29 Jul 2008, at 20:56, hari jayaram wrote:
>
> Sorry for that incomplete last post
>
> Thanks to everyone who wrote in with advice on scaling multi-crystal
> spacegroup P1 datasets together to increase redundancy for MAD/SAD/
> SAS phasing.
>
> I have three questions about scaling and merging multi crystal
> datasets in scala
>
> In our case we have the following
> 1) Crystal 1 - XNAME - C1 - DNAME peak1
> DNAME inf1
> 2)Crystal 2 - XNAME- C2 DNAME peak1
>
> The two crystals have cell dimensions within 1-2 % and I have cadded
> the output from mosflm for each of the crystals into one giant mtz
> file. No two batch numbers are the same .When we run scala the Job
> fails and We get errors like "Insufficient Data to determine
> parameters" erros - Too few reflections
>
> Of course scala on each crystal separately works great and gives us
> reasonable correlation coefficients ( 0.2 to 0.3 for Correlations
> within half dataset for peak and inflection to 3.2 A. The overall
> redundancy is around 6-8. Rpim around 0.05 )
>
> My question is can scala scale mutli-crystal datasets as I am
> attempting to do.
>
>
> 2) My second unrelated question is how can I DRENAME using cad on
> unmerged data. If I have run mosflm and then want to change the
> DNAME of a dataset , CAD says it can DRENAME a dataset , but at the
> same time CAD has a problem with unmerged mtz files. How can I edit
> the DNAME, XNAME, PNAME etc on an mtzfile output from mosflm
>
> Your help is greatly appreciated.
>
> Hari Jayaram
>
>
>
|