According to Engh and Huber the geometric restraints are meant to have
correct RELATIVE weights... I have forgotten how that was decided.
Eleanor
[EMAIL PROTECTED] wrote:
I do not think I made myself clear... WEIGH AUTO does not work very
well (and Clemens's attempt to apply a similar method to weigh the
X-ray terms in autoBUSTER) because all the types of geometrical
restraints (DIST, ANGL, PLAN, CHIR, NONB, etc...) are scaled by the
same factor for all resolutions, with rmsd for 1-2 bonds being used as
an indicator of how well we balance the X-ray terms with the
geometrical ones (in the parameter matrix).
This indeed can lead to larger bond rmsd for high-resolution
structures (competition between DIST restraints with stronger X-ray
terms), and very small ones for low-resolution data (no competition,
DIST wins!).
I am suggesting applying different scale factors for different types
of geometrical restraints, depending on the resolution to have a more
realistic distribution of the deviations from "ideal" values for
different types of geometrical features.
Tadeusz
*"Roberto Steiner" <[EMAIL PROTECTED]>*
23-Aug-2006 12:05
To
[EMAIL PROTECTED]
cc
"CCP4 Bulletin Board" <[email protected]>, [EMAIL PROTECTED]
Subject
Re: [ccp4bb]: Refmac vs Ramachandran (low resolution)
Hi Tadeusz
Of course, one can adjust weights of individual types of restraints
(does anybody these days?), but It would be nice to have an automatic
weighing scheme in REFMAC, which would deal with the different needs
for restraints at different resolution ranges...
You can try
WEIG AUTO
The option is not yet available through the interface (I think). I
usually modify the CCP4i-generated refmac script with Run&View Com File.
Alternatively, you can just have a one-line file with the above
command and give the name of that file in the Developers option.
Roberto.
Tadeusz
*"P.R. Evans" <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>*>*
Sent by: [EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>
23-Aug-2006 10:17
To
"CCP4 Bulletin Board" <[EMAIL PROTECTED] <mailto:[email protected]>>
cc
Subject
Re: [ccp4bb]: Refmac vs Ramachandran (low resolution)
*** For details on how to be removed from this list visit the ***
*** CCP4 home page _http://www.ccp4.ac.uk_
<http://www.ccp4.ac.uk/> ***
When I've done searches of the CCDC small molecule database for molecular
fragments, the rmsd of bond lengths has come out in the range 0.02 -
0.03A.
These deviations between different structures are of course larger
than the
estimate rmsd within a single structure, which I think are the number
around 0.014A quoted earlier in this thread, and probably give a
reasonable
estimate of the uncertainty.
I usually try to adjust matrix weight to give rmsd bonds around 0.02A
Phil
---
Dr. Roberto Steiner
Randall Division of Cell and Molecular Biophysics
New Hunt's House
King's College London
Guy's Campus
London, SE1 1UL
Phone +44 (0)20-7848-8216
Fax +44 (0)20-7848-6435
e-mail [EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>
begin:vcard
fn:Eleanor Dodson
n:Dodson;Eleanor
email;internet:[EMAIL PROTECTED]
tel;work:+44 (0) 1904 328259
tel;fax:+44 (0) 1904 328266
tel;home:+44 (0) 1904 424449
version:2.1
end:vcard
