[phenixbb] Reproducibility of R-factors

[Nathaniel Echols]

On Tue, Sep 14, 2010 at 12:02 PM,  <miket at chem.ucla.edu> wrote:
> I recently used phenix.model_vs_data to evaluate a structure that was refined using Refmac. The R_work and R_free calculated by refmac are 0.18 and 0.20, while model_vs_data calculates them as 0.20 and 0.22. Should I be concerned about this discrepancy? Is it typical for different programs to calculate different values for R-factors using the same data, and if so what is the source of this difference? Seems to me like there shouldn't be much of a difference (at least not 2%). Maybe I'm wrong.

The structure was probably refined with "implicit" hydrogens in
Refmac, which aren't included in the PDB file.  model_vs_data won't
assume anything about atoms it can't see, and the absence of hydrogens
could easily lead to a 2% difference in R-factor at higher
resolutions.  (An excellent reason to always include hydrogens in your
PDB depositions!)

-Nat