[sharp-discuss] Sushi on Linux box/SGI

[Clemens Vonrhein]

Dear Klaus,

> Changing the user status from beginner to expert does not solve the
> problem. I guess it might have to do with r/w status of the file that
> stores the preferences. Thus, which file stores the preferences?

A way to check if changes to your preferences are recognized correctly
by SHARP is to change something (e.g. switch from Standard to Expert
user, save it), going back to the main page and re-visit the
preferences page: it should now show the changed value. This means
that SHARP sees and understands your preferences (stored in
$BDG_home/users/klaus/sharp.dat - which needs only permissions to be
rw by the httpd process).

Another possible problem might be, if you have the caching policy of
your browser set to seomthing 'funny'. I usually set it to "Every
time" (for Netscape this is in Edit -> Preferences -> Advanced ->
Cache).

If the file doesn't show up: make sure it is called
something.extern.mtz, has permission 0644, contains columns HLA, HLB,
HLC and HLD and is in /where/ever/sharpfiles/datafiles. Now when
starting SHARP there should be a pull-down menu to select it.

If all this still doesn't work: right-mouse-click on the frame
containing the "Help", "New", "Delete2 etc. buttons and specify "View
Frame Source" - look for the definition of 'extern_abcd' near the top.

> In regards to refining/estimating the anisomorphism parameters: I was not
> able to get a stable refinement of the heavy atom parameters, unless I
> would allow to refine NISO_BGLO from reasonable estimates - reasonable
> because they roughly scaled with the Riso values and remained in the
> vicinity of 3 or below.

I would always refine non-isomorphism parameters - unless they go way
off sensible limits (e.g. > 100 for *_BGLO values). In cases with only
low-resolution data (below 3-3.5 A) and/or bad data one has to be a
bit cautious too. With the current version of SHARP I wouldn't bother
to estimate these values though: the estimates are sometimes not very
good and a few cycles of refinement _should_ get to decent values
anyway.

NOTE: estimation of scale factors at the beginning of your first
refinement is obviously a completely different matter.

> Obviously, I would not refine these parameters for the reference data set
> in an MIR case.

I would have been suprised about that anyway ... after all: the
reference dataset is exactly that: a reference for non-isomorphism and
scale, so these values can never be refined (although scale can be
estimated to get things on roughly absolute scale).

Cheers

Clemens

-- 

***************************************************************
* Clemens Vonrhein, Ph.D.          vonrhein@GlobalPhasing.com
*
*  Global Phasing Ltd.
*  Sheraton House, Castle Park     Tel: +44-(0)1223-353033
*  Cambridge CB3 0AX, UK           Fax: +44-(0)1223-366889
*--------------------------------------------------------------
* BUSTER Development Group      (http://www.globalphasing.com)
***************************************************************