Dear All,
I would like to announce the latest release of
xia2: Automated Data Reduction
xia2 is a new automated data reduction system designed to work from raw
diffraction data and a little metadata, and produce usefully reduced
data in a form suitable for immediately starting phasing and structure
solution, e.g. through Mr BUMP or your favourite experimental phasing
suite. The latest version fixes a number of significant bugs, so is well
worth the update.
The only software which is actually required is CCP4 6.0.1/2 (other data
reduction suites will be supported in time) and Python 2.4.x, and the
following platforms are supported:
- GNU/Linux on i386 (and 64 bit extensions)
- OS X 10.4 intel & PPC
- Microsoft Windows 2000/XP (Vista is untested)
For more information and downloads please visit the CCP4 prerelease
pages or go directly to:
http://www.ccp4.ac.uk/xia
Future release notices will be sent only to [log in to unmask] -
subscription details can be found on the above web page. Release notes,
changes and bug fixes since previous versions and platform specific
notes follow.
Regards,
Graeme Winter
-----------------------------
Development of xia2 is supported by the BBSRC e-HTPX grant and the EU
framework 6 BioXHit initiative. xia2 makes extensive use of Phil Evans's
program Pointless, and I would like to thank him for the extensive
modifications made to the program, and can will also use Labelit if
available. I would also like to thank the US structural genomics JCSG
group for providing their extensive collection of raw diffraction data
to methods developers. Finally I would like to acknowledge CCP4, for
without their software this project would go nowhere!
Release Notes
-------------
Changes since 0.2.4:
- xia2setup will now add the f', f'' values if a scan is available and
has been processed by chooch.
- GAIN now estimated by diffdump - enabling future parallelisation of
the integration stage.
- For detectors formed as a mosaic of a number of tiles, with "gaps" in
between, now mask those gapped areas in Mosflm to reduce the number of
bad reflections.
- Now has a fix for the ongoing indexing problem which gave rise to
errors like "something horrible has happened in indexing".
- The "best" log file for each process will now be recorded in a
LogFiles directory - which means no more hunting around for the final
scala log file and what not.
- Scaling now produces unmerged reflection files in scalepack format as
well as merged scalepack and merged MTZ format. This is done by scala
(if used) by recycling the final SCALES and adding "output polish
unmerged."
- Now runs a twinning test using the CCP4 program sfcheck, and will warn
you if your data look twinned.
- Now allow up to 100 cycles of scale refinement to cope with more
extreme cases where data at very different resolutions are scaled
together.
Changes since 0.2.3
- Added preferences file
- Added xia2setup program to create the .xinfo file - this will also run
LABELIT to configure the beam centre if it is installed
- Added run-time check that CCP4 and so on are available
Changes since 0.2.2.4:
- Changed to a BSD license
- Added "python setup" check to xia2
- Relaxed criteria on isomorphism to 1% not 0.5A etc.
- Now wavelengths in wavelength record will override header values if
provided, but do not (indeed, should not) be included if wavelength
values are correct.
Changes since 0.2.2.3:
- Created version for Power PC G4/G5
- Fixed .csh setup scripts
- If wavelength values not specified in the .xinfo file will use values
from the image headers
- If loggraph output from scala is broken by having Mn(I/sigma) greater
than 99.99 can now cope (ignores that record)
- Will now use a maximum of 180 degrees of data for deciding e.g. the
point group, spacegroup and correct indexing standards - this helps the
cases where exceptionally high redundancy data has been measured
Changes since 0.2.2.2:
- Updated printheader to cope with MAR 165
- Cleaner error messages when data missing
- Added support for MAR image plate data
- Included more example .xinfo files in the examples directory
- Relaxed refinement parameters in Mosflm cell refinement
Platform Specific Advice
------------------------
Gfortran compiled CCP4
Recent versions of the gcc compiler suite have included a new compiler
called "gfortran" in place of "g77". This has, unfortunately, been
designed with optimisation too high on the agenda, and the output from
programs can sometimes be mashed. If this happens you need to set
"GFOTRAN_UNBUFFERED_ALL" to 1 in your environment.
Linux
There should be no particular problems on this platform - assuming that
you have a recent version of Python. Any 2.4 version should work, but I
have mostly used 2.4.3 and 2.4.4, and would recommend them. If the
version you have on your system by default (type "python" to find out
what version you have)is too old, you can easily install a new version,
perhaps in the xia directory, from here. This is best installed by
unpacking the tarball (tar xvfz Python-2.4.4.tgz) then doing ./configure
/where/I/want/it then make, make install.
OS X 10.4
The only potential problem with this platform is that the Python which
comes as standard is 2.3 - however a universal binary is available from
here, which should sort out any problems. Just follow the instructions
in the disk image.
Windows XP
Once more you will need a recent version of Python - which can be found
here, however, you will also need "pywin32" to provide a little more job
control functionality - this is available from here. Both are simple,
double-click, binary installers. There is now a version of Labelit
available for Windows, which can be found here.
The mosflm binary which comes with iMosflm works well with xia2. You
will however need to register as beta tester of iMosflm to obtain this.
|