Dear Al,
Refmac by default divides reflexions in 20 shells ("bins") - this can
mean a few of them do not contain a thin Rfree shell.
When this happens I add "bins 10" to the refmac com-file, or, in
CCP4i, instead of "run", do "run and view com file" and add "bins 10"
to the file before continuing. This usually resolves the issue.
Mark
Quoting "Alasdair K. Mackenzie" <[log in to unmask]>:
> Greetings CCP4-ers,
>
> The situation:
> I have 10 fold NCS, and thought it would be fun to try the Rfree in thin
> shells.
> After using SFTOOLS (rfree 0.05 shell) to divvy up the Rfree flag,
> everything went swimmingly until i looked at my REFMAC log in the section
> "Things for loggraph, R factor and others". There in the table for R/Rfree
> vs resolution there were a few ******'s, which were apparently caused due
> to missing Rfree flags in the various REFMAC determined resolution shells.
>
> When i add up the number of reflections set aside for Rfree in the table i
> get 5%, so it appears to me as if REFMAC is using all the Rfree flags
> anyway, but just using its own definition of shells for resolution stats.
>
> The question is:
> Are the Rfree stats OK, or do i need to try and match the resolution
> shells for Rfree flags with the resolution shells that REFMAC is using?
> (if the latter holds true can someone give me a clue on how to do that?)
>
> cheers,
> Al
>
|