[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ccp4bb]: New CNS compatible version of DDQ structure validation program

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

Dear all,

I would like to bring to your attention a new version of the
crystal structure validation program called DDQ (van den Akker & Hol. 
Acta D55: 206-218, 1999). After the previous version I received several 
suggestions of making DDQ compatible with CNS which I am happy to say is 
now the case.

FYI, DDQ is, as far as I know, the only validation program which ranks 
the accuracy of your structure against a large number of randomly 
selected structures from the PDB in a resolution dependent manner 
(ranked as either worst, bottom 25%, below average, above average, top 25%,
and best). I think this is important since there currently appears to be
no consensus on what R and Rfree values should be for a finished crystal
structure. Even worse, the R, R-free and stereochemistry based validation 
methods, such as PROCHECK, can be (deliberately or not)
doctered/biased due to your choice of refinement protocols (use a
low data-parameter ratio and then change your R-free set of reflections
at the very end of refinement and you will get a 'nice' low R *and* Rfree). 
DDQ scores however cannot be forged to my knowledge as it quantifies an 
unbiased signature of correctly and completely
refined crystal structures and that is: very strong and correctly 
positioned positive water peaks in a hydrated difference map (i.e. 
waters deliberately removed from map calculation) and absence of shift 
peaks and UFO peaks (due to unmodeled residues, ligands, etc.). 
The latter is unfortunately a forgotten but important aspect in  
validating structures (because a partial correct structure is still 
partially incorrect by omission). 

DDQ has been tested thoroughly on >100 structures and could easily detect 
10 unnoticed errors in a published 2.4 Angstrom resolution structure 
(R<20%) with no false positives (including a peptide flip, frame shift 
etc.) as well as errors in high resolution structures (See Acta D article).

DDQ combines both global accuracy (i.e. low phase error --> strong
water peaks) with local accuracy (correctly positioned water peaks, no 
shift peaks) to arrive at local and global quality indicators. 
The DDQ program can also aid in speeding up refinement and also be used 
as a quantitative difference map quality indicator (for example, when 
you need to play with different scales or resolution ranges etc.).

For more information, to obtain a PDF file of the DDQ Acta D paper, 
or to download the (free) DDQ program please visit our website:
Focco van den Akker

Dr. F. van den Akker
Department of Molecular Biology/NB-20 
Cleveland Clinic Foundation
Lerner Research Institute
9500 Euclid Av.                        Tel: 216-444-3588/Fax: 216-445-1466
Cleveland OH 44195                     e-mail: focco@pierre.lerner.ccf.org
              Website: http://www.lerner.ccf.org/labs/yee