[phenixbb] Still can't get refinement of anomalous groups to work

[Nathaniel Echols]

On Wed, Jan 16, 2013 at 2:34 PM, Brennan Bonnet
<Brennan.Bonnet at lightsource.ca> wrote:
> I'm still working on this pesky anomalous group refinement problem.  Originally, the problem was that the output from AutoBuild no longer retained the anomalous data.  To fix this, I used the Phenix reflection file editor and combined exptl_fobs_phases_freeR_flags.mtz (which is automatically entered as input when running phenix.refine from AutoBuild) with the F+ F- values from my original reflection data.  Now when I try to run refinement I get this error message:
>
> "This model appears to have previously been refined against a different set of R-free flags.  Please resolve the mismatch."
>
> I find this strange since I am only modifying the file that phenix.refine tries to use anyway.  How do I fix this problem?

There are several different reasons why this could happen, mostly
having to do with the fact that the reflection file editor will extend
your flags to cover all reflections.  My guess is that the exact set
of reflections changes slightly when you transfer the anomalous data,
and the flags are therefore different.  (I don't think you're doing
anything wrong - it is very easy to get this error message
accidentally, but it also catches some genuine mistakes, so it's quite
valuable.)  Just click the "Options" button in the "X-ray data and
experimental phases" section, set the user level to "Advanced", and
look for the control labeled "Ignore PDB hexdigest sanity check".

This probably belongs in the FAQs...

-Nat