Dear Almudena,

you have to consider what type of questions you want to answer from your PDB 
file - the term 'reliable' has to be put into context.

The level of detail you could derive from a 2A resolution probably does not 
depend much on the coordinate uncertainty, so I dare say both models are 
equally reliable for common applications (assuming validation procedures have 
been taken into account at common sense).

The coordinate uncertainty also depends on the uncertainty of the cell 
dimensions, and as far as I know these are not recorded in a PDB or MTZ file.

Best regards,
Tim

On Tuesday 28 February 2017 02:50:01 PM Almudena Ponce Salvatierra wrote:
> Dear all,
> 
> I would like to know how the estimated coordinate error is calculated in
> phenix.refine, the one that is written on the pdb file as "ML based".
> 
> I'm asking due to the following:
> 
> For example if I am comparing two different models (structures determined
> at 2.1 and 2 Angstroms) but one does have a coordinate error estimate of
> 0.25 Angstroms and the other one of 0.6, respectively. Is the first model
> more reliable? The second one is reliable too but my question points
> towards how conservative or how more conservative should I be when
> describing distances , potential H bonds, coordination. Despite those
> different numbers in estimated coordinate error, the electron density maps
> look very similar or almost the same for both structures, very clear and
> complete at 2 Angstroms. That's why I was somehow surprised to find these
> difference.
> 
> Thank you very much in advance
> 
> Best
> 
> Almudena
> 
> 
> 
> --
> Dr. Almudena Ponce-Salvatierra
> EMBL-European Molecular Biology Laboratory
> 71, avenue des Martyrs - CS 90181
> F - 38042 Grenoble Cedex 9
-- 
--
Paul Scherrer Institut
Dr. Tim Gruene
- persoenlich -
Principal Investigator
Biology and Chemistry
OFLC/104
CH-5232 Villigen PSI

Phone: +41 (0)56 310 5297

GPG Key ID = A46BEE1A