shelxD solution criteria

From: Andre Godoy
Date: 11 November 2011 11:31

Hello everyone.

I'm a begginer in crystallography, especially in EP and I'm trying to solve a structure by SAD. Right now I'm trying to solve the substructure on shelxD by running hkl2map, but I'm finding hard to chose between solutions. So I'm asking, what should by my solution criteria (CCall, number of diferent solutions, site occupancy, or something else...). Also, how I can choose the HA sites number and resolution cut-off. Many "tutorials" have different awnsers for those question, so I'm a little lost right now. Any help will be apreciated!!

Best wishes to all

Andre Godoy

----------
From: Tim Gruene 

Dear Andre,

as a rule of thumb you should cut the resolution for shelxd in SAD where
the anomalous selfCC as reported by shelxc drops below 30%. The second
important figure is a correct solvent content for shelxe, but only
really if you do not use the current beta version of shelxe (available
upon email request to George Sheldrick or me) which is a lot less
sensitive to this parameter thanks to its autotracing.

Pick the correct solution by looking at the phs-files created by shelxe.
In my opinion this is much better (at least for "visual" people) than
looking at figures and statistics.

The phs-files can be loaded into coot after loading the .hat-files also
created by shelxe (if coot crashes when you load the hat-file, edit it
and remove everything below the first occurrence of the word "END").

Tim

- --
- --
Dr Tim Gruene
Institut fuer anorganische Chemie


----------
From: Eleanor Dodson

If you are lucky, and run SHELXE after SHELXD then the hand with the best CC is probably the correct one.
The string SHELXC/D/E are automatically run together from the CCP4 GUI, and if you look at the loggraph output you will see the contrast (hopefully!)

Eleanor

----------
From: Tim Gruene


Hi Paul,

I just checked and did notice that coot (version 0.6.2) does not crash
anymore in the presence of two 'END' cards in the .hat-file.

I am sorry for mis-blaming coot.

Cheers,
Tim

On 11/14/2011 04:39 PM, Paul Emsley wrote:
> Do you know what makes coot crash?
>
> Is this in old versions of coot?
>
> Paul.