John R. Horton wrote:
> I've tried several ways to get an Rfree flag into my mtz
> file...converting from a CNS to mtz (with existing flags) and scalepack
> to mtz and etc etc...with the same result, i.e. that the "bin number"
> that the reflection is in (0-19) is put in mtz file as the Rfree flag
> and not 0 or 1. I was wondering if this has to do with array size or
> something. This is a 1.42A dataset with 125982 reflections.
In MTZ files, free-R flags are usually from 0-19 (or 0-N, if you are not
using a 5% free set).
There is a good reason for this - if you are not sure whether your FreeR
is meaningful or not, you can re-run your structure solution using a
different, non-overlapping FreeR set just by telling each individual
program to use a set other than 0 as the test set. By this means you can
do a full cross-validation.
Has anyone ever used this in practice? I don't know. But it is a good
design feature.
|