To try to answer the Q I think you are asking..
If you keep anomalous seperate you will get a file from ctruncate with h
k l F+ SIGF+ F- SIGF- I+ SIGI+ I- SIGI-
The observations flagged as F- or I- etc are actually measured for the
reflection -h-k-l
So uniqueify generates markers for each possible h k l (but not for
-h-k -l) to a given resolution limit and outputs a file h k l Marker
Then cad combines that file with the ctruncate output and if requested
also picks up a FreeRflag generated for another data set.
for that entry so both hkl, and -h -k -l will belong either to the Free
or the working set.
freerflag is a tool which generates Free flags for a requested % of this
data set. If some Free flags are already assigned it a) guesses the %
chosen for that set b) generates FreeR flags for the same % but only
assigns ones to those h k l which do not aready carry a FreeR flag.
Is that clear!! It is a bit complex..
Gets even more so if you want to consider twinning.. In that case the
FreeR flags must be generated in the highest possible pointgroup and
extended in cad to the actual one, This means that twinned pairs also
belong either to the Free or the working set.
Eleanor
# On 10/13/2010 08:07 PM, Lepore, Bryan wrote:
[ ccp4 6.1.13 ]
[ gui 2.0.6 ]
if a data set with merged Bijvoet pairs and a free set is then "made anomalous"
- i.e. re-scaled with separated Bijvoet pairs - i do not understand how acentric
reflections in the free set are extended by cad or freerflag or uniqueify.
the easy question : is [uniqueify with extension] ->[cad with combination but
not extension] adequate?
i.e. i gather that cad preserves FreeR records between two resolution limits
(e.g. 500-3.5) and only extends them in the higher resolution segment (e.g. 3.5
- 2.1).
so do i have this right : for acentrics, -h-k-l will be preserved while hkl go
to the work set. That will decrease the fraction of reflections in the Free set
by half the number of acentrics in the merged free set, because they would be
in the work set. centrics do not meet this problem.
-bryan
