[phenixbb] R value difference in model vs data vs refinement

[Pavel Afonine]

Hi Christian,

I keep seeing people saying this but I've never seen a case where 
R-factors reported by phenix.refine and phenix.model_vs_data are 
different given identical inputs. We even have a specific test that 
asserts that Rs out of both tools are identical (of course this test 
cannot cover all the variety of situations).

So if you send me the files that I can use to reproduce your observation 
that will help me to find what's going on.

Thanks,
Pavel


On 3/19/12 11:20 AM, Christian Roth wrote:
> Dear all,
>
> I refined a structure in phenix quite good resolution (1.4 Ang.)
> After the last refinement
> I created the "table 1" to extract all data. Interestingly I got a warning due
> to small changes in the R values
> R/Rfree after Refinement. 15,46/18,16
> Phenix modelvs.data gives me
> R/Rfree 15,34/18,07
> Interestingly this is nearly exactly the value which I got after the first OHS
> scattering optimization from phenix
> R/Rfree  15,43/18,03
> The differences are of course very small and negeligible, but  I wonder why it
> happens. Esspecially taking into account that it was refined with phenix and in
> this case I would expect that both values match exactly.
>
> Thank you very much in advance
>
> Best Regards
>
> Christian