*** For details on how to be removed from this list visit the *** *** CCP4 home page http://www.ccp4.ac.uk ***
> Now if you approximate a TLS model in terms of individual ellipsoids, > which is what e.g. refmac does internally during refinement, the > same issue would arise when applying bond-length restraints. > I do not recall whether refmac does or does not correct for the > apparent compression in this case. Perhaps Garib will speak up. It doesn't, unless Garib has changed things more recently. The correction depends on the magnitude of L (see section 2.1 of our 2001 TLS paper, Acta Cryst D57, 122, and references therein). For domain-sized TLS groups, L tends to be smaller, and the correction is insignificant. It could become significant for small TLS groups, e.g. of a few residues only, or side chain groups. The compression arises from the shift of the observed mean position with respect to the rest position. The estimate of this shift depends on the TLS model used. This is partly a consequence of the TLS formalism, but also the TLS model used will affect the estimate of the mean position - the actual x, y, z you record. Martyn
