*** For details on how to be removed from this list visit the *** *** CCP4 home page http://www.ccp4.ac.uk ***
This alternative indexing problem is addressed by a new program I have been working on, Pointless. This program also can be used to determine the Laue group & possibly the spacegroup of unmerged data. There is a beta release of this available as Linux & OSX executables by anonymous ftp at ftp://ftp.mrc-lmb.cam.ac.uk/pub/pre/pointless-1.0.3.linux pointless-1.0.3.osxppc ... pointless.html # documentation Note that various bugs have been fixed relative to earlier versions. Other executables could be made available on request. This program will be in the next CCP4 release (DV) Given a merged reference file in the indexing that you want reference_data.mtz, and a test data file (merged or unmerged) test_data.mtz, then the program can be run as pointless hklin test_data.mtz hklref reference_data.mtz hklout newdata.mtz and the new file will contain a copy of the hklin file reindexed to match the reference. This will cope with accidental indexing alternatives (eg a~=b in orthorhombic) as well as exact ones as in P3 etc I'd welcome any feedback from anyone who tries this program Phil > > > Hi Ron: > > I think my previous response was worded quite poorly. The essence > of the problem is that a != b in a trigonal system (unlike > hexagonal). So if you have two data sets (say a native and a > derivative) and you process each of them, there is a priori a 50% > chance that what you called "a*" for the native will be "b*" for the > derivative. Neither is really right or wrong. It is only relative to > an arbitrary reference. > > Since a and b are not equivalent, it isn't simply a rotation, it > really is a reindexing. Riso for a good derivative might be about > 15% when they are both indexed the same way, and shoots up to 40% or > more if indexed the "wrong" (relatively incompatible) way. > > Similarly with molecular replacement, so if you are doing molecular > replacement, for example, the Fcalcs generated by ccp4 or CNS have a > 50% chance of choosing "a" differently from your native data set. > > I've done a lot of work in that space group in the last 12 (yikes) > years. So I've become painfully familiar with this. > > There is a discussion in the CCP4 users guide in the $CCP4/manual > directory that describes this much better than I have. > > HTH, > > Bill > > > > > > On Mar 28, 2006, at 10:12 AM, R. Stenkamp wrote: > >> Hi. >> >> Your statement about the indexing issue for trigonal space groups >> confused me. I've not seen ccp4 programs change the indices (unless >> explicitly instructed to do so). Isn't the choice of a* and b* made >> back at data processing/indexing time? >> >> I know the indexing ambiguity makes a lot of difference when >> comparing data sets, even MAD sets at different wavelengths, but >> how would an indexing issue affect a molecular replacement and >> refinement problem? Re-indexing of hkl corresponds to a rotation >> of the reciprocal lattice, so the structure ought to be just as >> readily described for the reindexed data as for the original >> setting, right? Again, the problem comes when you try to compare >> two structures or data sets that may differ by the rotation. >> >> >> Thanks. >> >> Ron >> >> p.s. The a*-b* choice is also an issue in tetragonal space groups >> like P4(x). > >
