[phenixbb] structural validation and deposit format

[Nathaniel Echols]

On Dec 10, 2009, at 8:46 PM, rui wrote:
> Sorry for this stupid question. When I tried to use phenix to validate the structure, it generate geometry outliers, model validation, real-space correlations and a phenix_map_coeffs.mtz file which I assume could be open in coot. How should I use this generated mtz file? When I open in coot, it tellls me core dump.

This is almost certainly a problem with Coot - you should probably upgrade or reinstall it.

> I'm wondering if phenix will automatically do the corrections or should I look one by one in the log file and modify? It also generate a test_reduced.pdb, should I use this reduced pdb ( H added) for next round of refinement? Thanks.

We are working on tools for certain automatic corrections - for instance, the fix_rotamers option in phenix.refine, which will address both rotamer outliers and real-space fit - but it is a complicated problem.  You should still inspect everything manually.  Reduce will also automatically flip Asn, Gln, and His sidechains that are misfit, but this is not integrated very well with the rest of Phenix right now beyond reporting suggested flips in the validation GUI(s).  (This behavior is also disabled when running phenix.ready_set to add hydrogens.)

I would leave the file output by Reduce alone - it is required for the all-atom contacts analysis, but the rest of the validation code uses the original PDB file.  However, we do recommend that you refine with hydrogens, regardless of resolution.

> Is the format from phenix.refine ready for deposit?


I believe so, but Pavel may be able to clarify this.

--------------------
Nathaniel Echols
Lawrence Berkeley Lab
510-486-5136
NEchols at lbl.gov