[phenixbb] rfree flag fractions
Pavel Afonine
PAfonine at lbl.gov
Sat Feb 28 00:02:56 PST 2009
Hi Yusuf,
yes, you can request a desired fraction of 'test' reflections to be
generated (xray_data.r_free_flags.fraction=) and you can set up an upper
limit for the number of 'test' reflections.
The defaults are 0.1 and 2000, if I remember correctly, and as far as I
know they are good in most of cases. The free-R flags should be evenly
distributed over the whole resolution range and each "narrow enough"
resolution bin should contain not less than 50 'test' reflections so the
ML estimates of distribution parameters (alpha/beta or sigmaa) are
reliable. The default in phenix.refine is set up to 140 'test'
reflections per bin which is found by systematic search.
I can imagine that one needs to lower these values if there are too few
data available so setting even a small fraction of reflections aside may
be critical. I hope this is not your case at 3A resolution and 12680
reflections!
Pavel.
On 2/27/09 11:58 AM, Yusuf Akhter wrote:
> Hi Everybody!
>
> I am using phenix version 1.3b for refining a structure at 3 A
> resolution and total number of total reflections are 12680.
>
>
> I have seen a provision in phenix to change the default maximal number
> of test flags to be generated and the fraction with this command line
>
> xray_data.r_free_flags.fraction=0.05 xray_data.r_free_flags.max_free=500
>
>
>
> My queries are
>
> 1)when should somebody use this utility ??
> 2)what is minimum number of Rfree flags one should use on this resolution and this total number of reflections??
>
>
> Any kind of suggestions will be appreciated.
>
>
>
> Cheers!
> Yusuf
>
>
>
>
>
>
>
>
>
>
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://www.phenix-online.org/mailman/listinfo/phenixbb
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20090228/56805a85/attachment-0003.htm>
More information about the phenixbb
mailing list