***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***


I repeated a couple of tests.  I had had a typo in one of the scripts.
I can confirm Johan's findings that having a valid "reference film" 
appears to be the critical parameter as to whether or not the postrefined
values get placed in the .sca header or not.  

Sorry for my incorrect conclusion about the fit cell vs fit a* b* c*, this
does not appear to be important.  Adding a valid reference film command
to my standard script resulted in the postrefined values getting placed
in the .sca header.

Regards,
Mitch


-----Original Message-----
From: Johan Turkenburg [mailto:[EMAIL PROTECTED] 
Sent: Wednesday, January 11, 2006 7:15 AM
To: Miller, Mitchell D.
Cc: Phil Jeffrey; Edward Berry; Alessio Accardi; Vaheh Oganesyan; 
[email protected]
Subject: Re: [ccp4bb]: Unit Cell Parameters


Also done a few more tests:

In my hands, changing from 'fit crystal cell' to 'fit crystal a*' and 
'fit crystal c*' (for a hexagonal crystal) makes no difference: in both 
cases the cell in the .sca file is the cell from the postrefinement, not 
from the first .x file.

Changing the reference film from 1 to 5 to 10 makes no difference, the 
cell is from postrefinement, not from the .x file.

BUT, with no reference film in (i.e. taking out the 'reference film' 
keywords), the cell in the .sca file is the cell in the .x file for 
image 1 and NOT the cell from postrefinement.

The cell from the postrefinement is still the same as the cell coming 
out of the runs with a reference film keyword in (just as well!).

Not found another way of 'breaking' scalepack yet. By the way, this is 
all with 'format denzo_ip'. Not tried 'format denzo_york1'.

Johan


Miller, Mitchell D. wrote:
>   I looked at a test this morning before I sent my original post and
> observed the unit cell from the first .x file being placed in the 
> output .sca file header.  With so many counter examples, I went back 
> and ran some more tests.  It seems to depend on what keywords you use 
> whether or not the updated values are put in the .sca file.  The way we 
> typically run scalepack from the C-shell scripts, the postrefined cell values 
> do NOT make it into thee .sca file.  (This is with v1.98.3 on OSF1)
> 
>>From postrefinement
>  CELL   61.3725   62.6230   70.3258   90.0000   90.0000   90.0000      
> 270285.3
>  A  11001   61.3725    0.0032
>  B  11001   62.6230    0.0023
>  C  11001   70.3258    0.0036
>>From first .x file
>  unit cell   61.465   64.507   70.708   90.000   90.000   90.000
> from .sca file
>   61.465    64.507    70.708    90.000    90.000    90.000 p222 
> 
>   After looking at the scale.in file produced by HKL2000 (which
> does put the postrefined cell values in the .sca file). I found that
> there are several small changes that will "break" the undocumented
> feature Phil and Ed are relying on and one will end up with the cell 
> from the first .x file in the .sca file header.
> 
>   For example, the hkl2000 generated scale.in file uses an undocumented
> parameter for the fit command (fit cell). If I change this to 3 fit commands
> (for a* b* and c* as outlined in the manual) for an orthorhombic crystal, 
> the postrefined cell values were not be placed in the .sca file header.  
> Another 
> change that resulted in the postrefined cell not being added to the .sca file 
> header was to comment out the 'reference film' command.  Without this command
> the cell from the first .x file is placed in the .sca header.
> 
>   Simply changing my script file to include a reference film and use fit cell
> in place of fit a* fit b* and fit c* commands did not result in the 
> post-refined
> values being placed in the .sca file. So I suspect there are more combinations
> of keywords that will prevent the postrefined values from being placed in
> the .sca header.
> 
>   Based on my tests and Alessio's original post, I would caution that one 
> check the header to confirm that they are getting the post-refined values
> if they use their own scalepack scripts or edit the scalepack input files 
> from HKL2000.
> 
> Regards,
> Mitch
> 
> 
> 
> -----Original Message-----
> From: Phil Jeffrey [mailto:[EMAIL PROTECTED] 
> Sent: Tuesday, January 10, 2006 8:25 AM
> To: Miller, Mitchell D.
> Cc: Edward Berry; Alessio Accardi; [email protected]
> Subject: Re: [ccp4bb]: Unit Cell Parameters
> 
> Miller, Mitchell D. wrote:
> 
>>The behavior you attribute to early versions of scalepack is still
>>described in the current manual.
> 
> 
> Which assumes that they updated the manual when they updated the 
> software.  I've seen the same behavior that Ed Berry reported.
> 
> Here is an example with 1.97.2 on Linux, run again just now to check:
> 
> Cell from .sca file:
> 125.486   125.486    59.256    90.000    90.000    90.000
> 
> Cell from first .x (denzo output) file:
> 126.422  126.422   59.575   90.000   90.000   90.000
> 
> Cell from postrefinement in scalepack:
> 125.486 125.486  59.256  90.000  90.000  90.000
> 
> It would be very interesting to hear from anyone running 1.97 or later 
> that did NOT see this behavior, since like Ed I've come to rely 
> on/expect this new behavior.
> 
> Regards,
> Phil Jeffrey
> Princeton
> 
> 
> 

-- 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Dr. Johan P. Turkenburg                     X-ray facilities manager
York Structural Biology Laboratory      
Chemistry Department                        Phone (+) 44 1904 328253
University of York                          Fax   (+) 44 1904 328266
York YO10 5YW   UK                          Email: [EMAIL PROTECTED]
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Reply via email to