Could this be a long-standing bug in the ccp4i Refmac GUI? Some time
ago, I sent a bug report to the ccp4 developers: if you did TLS
refinement, you got an additional tab where you should set the B-factor
to a uniform value. In the next round of refinement, where you read in
the refined TLS parameters from the previous run, this tab vanished, but
the B-factor was nevertheless still set to the same uniform value, which
resulted in high starting R-factors for the first cycle!
I haven't tested this recently, but if you display the actual Refmac
command script from the GUI, it should be obvious whether this bug is
still there or not.
Best regards,
Dirk.
Am 25.02.10 17:54, schrieb Jan Schoepe:
What I tried for instance was in refmac:
1.) 20 cycles TLS followed by 10 cycles restrained refinement which
gave Rwork=24%/Rfree=33%.
Afterwards I did 0-1 cycles restrained refinement with output from
prev. run (pdb file, mtz file, tls file) as input for that. Result:
33%/44%.
2.) I omitted the restrained refinement, else same as above. Result
was 26%/32% after TLS and 31%/38% after restrained refinement...
--- gbirr...@bidmc.harvard.edu /<gbirr...@bidmc.harvard.edu>/ schrieb
am *Do, 25.2.2010:
*
*
Von: gbirr...@bidmc.harvard.edu <gbirr...@bidmc.harvard.edu>
Betreff: RE: [ccp4bb] about comparing Rwork and Rfree factors of
different refinement trials
An: j.scho...@yahoo.de
Datum: Donnerstag, 25. Februar, 2010 17:12 Uhr
*
* *
*You can fix the TLS parameters as well. *
*Zero cycles of Refmac and fixed TLS*
*Will this work in your case?*
* *
* *
* *
*Gabriel Birrane**
Beth Israel Deaconess Medical Center
RN348
99 Brookline Ave
Boston, MA 02215 *
* *
*
*
------------------------------------------------------------------------
**From:* Jan Schoepe [mailto:j.scho...@yahoo.de]
*Sent:* Thursday, February 25, 2010 11:06 AM
*To:* Birrane,Gabriel (Bidmc-Research Fellow)
*Subject:* RE: [ccp4bb] about comparing Rwork and Rfree factors of
different refinement trials
*
In principle yes, but I doubt that this works properly, especially
after TLS refinement.
--- gbirr...@bidmc.harvard.edu /<gbirr...@bidmc.harvard.edu>/
schrieb am *Do, 25.2.2010:
*
*
Von: gbirr...@bidmc.harvard.edu <gbirr...@bidmc.harvard.edu>
Betreff: RE: [ccp4bb] about comparing Rwork and Rfree factors
of different refinement trials
An: j.scho...@yahoo.de
Datum: Donnerstag, 25. Februar, 2010 16:04 Uhr
*
*
You can do zero cycles of refinement in Refmac. This will just
give you the statistics as Refmac calculates them without
any refinement. Is this what you are looking for?
________________________________________
From: CCP4 bulletin board [ccp...@jiscmail.ac.uk] On Behalf Of
Jan Schoepe [j.scho...@yahoo.de]
Sent: Thursday, February 25, 2010 9:38 AM
To: CCP4BB@JISCMAIL.AC.UK
Subject: [ccp4bb] about comparing Rwork and Rfree factors of
different refinement trials
Dear all,
I do have a question about comparing Rfree and Rwork factors
of different refinement trials whereas I always started with
the same pdb file and structure factors (phasing by MR).
Means I had a protein structure which was (not just by me)
refined several times in different ways also with different
programs and also accordingly got different R factors for each
finally refined structure.
Could anyone suggest if there is something like a "standard
run" which makes all these R factors (or "derivatives" since
this run should change the R factors) better comparable?
(E.g. it did not work for me to do a 1 cycle rigid body
refinement in refmac hoping that the R factors are measured
well and the structure does not change much. In fact, the R
factors increased dramatically, lets say Rfree from 30% to 40%.)
Many thanks for your suggestions!
Jan
__________________________________________________
Do You Yahoo!?
Sie sind Spam leid? Yahoo! Mail verfügt über einen
herausragenden Schutz gegen Massenmails.
http://mail.yahoo.com
*
*
__________________________________________________
Do You Yahoo!?
Sie sind Spam leid? Yahoo! Mail verfügt über einen herausragenden
Schutz gegen Massenmails.
http://mail.yahoo.com *
__________________________________________________
Do You Yahoo!?
Sie sind Spam leid? Yahoo! Mail verfügt über einen herausragenden
Schutz gegen Massenmails.
http://mail.yahoo.com
--
*******************************************************
Dirk Kostrewa
Gene Center, A 5.07
Ludwig-Maximilians-University
Feodor-Lynen-Str. 25
81377 Munich
Germany
Phone: +49-89-2180-76845
Fax: +49-89-2180-76999
E-mail: kostr...@genzentrum.lmu.de
WWW: www.genzentrum.lmu.de
*******************************************************
