[sharp-discuss] output of autoSHARP

[Clemens Vonrhein]

Hi Jie,

On Thu, Apr 06, 2006 at 10:46:13AM -0700, Jie Li wrote:
> Hi, everyone,
> 
> I ran the autoSHARP using some HA positions (totally 34 HA  
> positions)  from SHELX.  The output from the autoSHARP shows that  
> 0/34 positions are found.

This sounds weird - when do you get this message? Make sure that
autoSHARP reads your input HA positions correctly, i.e. the format of
your xyz.hatom file needs to be correct:

 - either use fractional coordinates

 - or a PDB file WITH THE CORRECT ATOM NAMES! (SHELXD writes out PDB
   files with sometimes wrong atom names)

Obviously, you could also use autoSHARP for all steps, including heavy
atom detection (so no need for running SHELXC/D outside): the latest
release will use SHELXC/D anyway for finding sites. This way you are
100% sure that everything will be parsed along correctly.

> Then SHARP refined these HA positions automatically. The final FOM
> is very low, about 0.1.

I guess this is the overal FOM, right? More interesting statisitcs are
Phasing Power against resolution: you expect good phasing power (>
1.0) at lower resolution and lower phasing power at high
resolution. At what resolution does the phasing power drop below 1.0?
This will tell you up to which resolution there is real signal in the
data.

> Does these mean these HA positions are not correct?

Depends.

> In addition, I found that SHARP always generate apparently perfect HA  
> map calculated from resid.mtz whatever HA positions were put in. How  
> could I judge the results?

The resid.mtz file contains all kind of columns for calculating maps:
the HAT maps are the least interesting ones: it just shows you your
current model. The important ones are the LLG (residual) maps. I
_highly_ recommend looking at these maps through the httpd interface:
it has several very easy to use tools for analysing these maps.

What you want are 'clean' LLG maps:

 - no unexplained positive peaks (additional sites)

 - no positive/negative peaks _at_ the HA positions in ANO maps (wrong
   f")

 - no positive/negative features around HA sites (anisotropic B)

> Is the density map from eden.mtz is the only way to judge it? (The
> density map is so bad that I can hardly judge the map quality).

These are the pure Heavy-atom phasing maps: hardly anyone is looking
at these. You should look at some density modified maps.

When you run autoSHARP I hope you specified to run all the way down to
(including) density modification, right? There you should get two
scores for the two hands (enantiomorphs): do you get a 'significant'
difference in these scores? I.e. does it point stongly to one of the
two hands?

If not: then your HA model at that stage is most likely wrong. If yes:
then you're on the right track (even if the maps aren't good).

If you specify to run autoSHARP all the way down to automatic building
(also recommended), autoSHARP will use ARP/wARP to either build a
complete model (if resolution is good enough) or to at least build
some helices (at lower resolution). This will also tell you how you're
doing.

 ----------------------------------------------------------------

Summary:

 - let autoSHARP try to find the HA sites (using SHELXC/D)

 - tell autoSHARP to run all the way down to automatic building


Cheers

Clemens


> 
> Thanks.
> 
> Jie
> 
> 
> _______________________________________________
> sharp-discuss mailing list
> sharp-discuss at globalphasing.com
> http://www.globalphasing.com/mailman/listinfo/sharp-discuss
> 

-- 

***************************************************************
* Clemens Vonrhein, Ph.D.     vonrhein AT GlobalPhasing DOT com
*
*  Global Phasing Ltd.
*  Sheraton House, Castle Park 
*  Cambridge CB3 0AX, UK
*--------------------------------------------------------------
* BUSTER Development Group      (http://www.globalphasing.com)
***************************************************************