[phenixbb] clashscore discrepancies
Alex Theodossis
alex.theodossis at monash.edu
Wed Apr 24 20:38:16 PDT 2013
Dear phenixbb,
Have refined a 1.05A structure with riding hydrogen model in phenix
v1.8.1-1168 and doing final validation.
The molprobity statistics output by phenix include:
all atom clashscore = 4.12
Rotamer outliers = 0.71%
Overall score = 1.2
Put through the molprobity server I am given the option of analysing with
existing hydrogen model, or replacing with the 'electron-cloud parameters'
they recommend for X-ray crystal structures. The results end up being very
different:
existing H model:
electron-cloud model:
All atom clash score: 4.66 (78th percentile) 3.72
(85th percentile)
Poor rotamers: 1.14%
1.14%
Overall score: 1.28 (88th percentile)
1.20 (92nd percentile)
While all three scenarios give acceptable statistics, I am surprised by the
inconsistency in the results given by molprobity, even when using the same
hydrogen model.
I am also wondering whether I am using a suboptimal riding model in my
refinement (generated using phenix ready-set).
Can anyone advise on the reason for these inconsistencies (is it just due
to differences in the version of molprobity being run in each case). Is
there anything I should be doing differently?
Thanks in advance.
On 18 November 2012 12:36, Alex Theodossis <alex.theodossis at monash.edu>wrote:
> Hi,
>
> I'm attempting to install the latest version of phenix on mac 10.5.8,
> using the command line installer.
>
> Installation proceeds well until towards the end, where it hangs on:
>
> precompiling .py files
>
> I can leave it for up to an hour without any further progress, but upon
> typing:
>
> ctrl+c
>
> The installation comes back to life and proceeds to completion.
>
> Can anyone advise as to what might be causing this and on how reliable
> this installation is going to be?
>
> Thanks in advance.
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20130425/324be3f2/attachment.htm>
More information about the phenixbb
mailing list