Dear Annette,
xprep (at least the xprep written by George Sheldrick, which comes with Bruker
machines) is not a data integration program!
Given that you collected data on a Bruker machine, presumably used saint/ the
apex software for data integration, then xprep, possibly after scaling the data
with sadabs, and finally pointless (which does - very roughly speaking - about
the same as xprep), chances are that your data were also run through scala and
maybe truncate.
This means that your data were scaled twice and maybe even truncated twice. In
the first case you only messed up your standard deviations, which would not have
such a severe impact on poor R/Rfree, but the latter case would explain the low
amplitudes. The default in scalepack2mtz, if I remember correctly, is to run
ctruncate, hence the amplitude.
Maybe you could carefully check all the steps you did. There is no need to run
pointless after converting the xprep output sca-file with scalepack2mtz.
Regard,
Tim
On Mon, Jun 25, 2012 at 09:25:23AM -0700, Annette Medina Morales wrote:
> Hi,
>
> We collected data using our in-house Bruker Proteum X8
> diffractometer and after using Xprep to integrate the images the
> output HKL file was converted to an mtz using CCP4's Scalepack2mtz.
> The prp file generated by Xprep shows good statistics with
> completeness and good data for 2.4 angstrom resolution. Mean
> intensity/sigma, Rsym, and Rsigma have good values. Pointless
> indicates space group is P212121. Later we used molecular
> replacement and obtained a good model and Refmac to refine the
> structure. The electron density maps show the correct density for
> the amino acids, waters, and other metals in the structure and
> refinement with coot improves the electron density around the atoms.
>
> However, after multiple rounds of refinement the R value drops to
> 0.25 but Rfree does not go below 0.35! Also, the structure factors
> are very low. We have tried all the obvious reasons as not having
> the correct space group which is not the case, and have included NCS
> restraints as well as TLS parameters to improve the refinement.
> None of this has helped in helping the Rfree value go down. Since we
> never have this problem with data collected from other detectors
> (like at SSRL), and this problem repeats with other datasets from
> collected in-house, we suspect that the original integration of the
> images using Xprep might be a problem and might be doing something
> to the reflection data. Does anyone have any experience with this?
>
> Also, I have recently tried using iMosflm to refine the data but
> using the .sfrm images from the Bruker detector is problematic since
> I get an Application error that reads "invalid command name" and an
> "Image read error" that reads "Error reading image header. Message
> from Mosflm is error reading record 33: check image is correct
> size". Does anyone have any advice on how to correct this since I
> understand that iMosflm is supposed to be able to read Bruker .sfrm
> images? Thanks! Annette
>
--
--
Dr Tim Gruene
Institut fuer anorganische Chemie
Tammannstr. 4
D-37077 Goettingen
GPG Key ID = A46BEE1A
|