Marilyn,
If you have post-refined the data (likely; default in HKL2000 & mosflm), then you should have standard deviations reported for the cell axes, e.g. a = 51.145 +/- 0.125 Angstrom. Since the PDB doesn't have a way (that I know of) to report the SD, then it boils down to which is better to report: 51.145 or 51.100. I think the former is more likely to be correct, but perhaps someone more versed in the statistics of data processing could comment.
Small molecule crystallographers often report 3 digits after the decimal point; in their case, this practice is likely more justified, due to the measurement of very high 2-theta angle reflections. On the other hand, postrefinement with low mosaicity crystals (e.g. many viruses) is a very powerful tool to give accurate cell constants with low % error.
Bottom line, however, is as both Ed & Victor mention: typical limiting factors are wavelength accuracy (and wavelength bandpass), optics (which define divergence), crystal mosaicity, etc. (Latter two are in theory decoupled from the wavelength/cell length dependency during postrefinement, but in practice they can influence the refined values). Level of agreement from different swaths of data can be very revealing
Dave
~~~~~~~~~~~~~~~~~~~~~~~~
David W. Borhani, Ph.D.
Structural Biology Group Leader
Chemistry Department
Abbott Bioresearch Center
Vox: 508-688-3944
Fax: 508-754-7784
Email: [EMAIL PROTECTED]
Smail: Abbott Bioresearch Center, Inc.
100 Research Drive
Worcester, MA 01605 U.S.A.
http://abbott.com/abbottbioresearch/
~~~~~~~~~~~~~~~~~~~~~~~~
| Victor Lamzin <[EMAIL PROTECTED]>
Sent by: [EMAIL PROTECTED] 06/05/2006 02:01 PM |
|
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Dear Marilyn,
If you place 10 numbers after the decimal, the CRYST record will be more
precise but not necessarily more accurate. The accuracy of the value
51.1 is 0.05/51.1=0.1%. On a synchrotron beamline the accuracy in the
a,b,c parameters is mostly affected by the accuracy in the mean value of
the wavelength used. The latter can be around 0.01% but you should
probably check it with the beamlime responsible. Another factor
affecting a,b,c could be the distance to the detector (e.g. large
crystal of irregular shape and a short distance). There are many other
factors of hopefully lower significance.
With best regards,
Victor
> Hi all,
>
> I have a reviewer challenging the number of digits to the right of the
> decimal point that I have reported in a submitted manuscript. I had
> reported 3 digits to the right, frankly because that is what most of
> the programs I use report and the PDB coordinate file reports 3 digits
> to the right of the decimal in the CRYST1 record. The reviewer doubts
> that the value is accurate to that many digits, and suggested
> reporting to the tenth of an Angstrom. In corresponding with the PDB,
> they allow 3 digits to the right of the decimal, but even if one gives
> the PDB, say 51.1 A as an axis, the CRYST1 record will be padded and
> will appear as 51.100A.
>
> So my question is .... how does one determine what is a valid accuracy
> or precision in unit cell axes? To be honest, I'm not sure what all
> the numbers are, and how many significant digits they have, that are
> used in calculating unit cell axes.
>
> In my particular case, I have data to 1.73 A, synchrotron radiation,
> 0.41 degree mosacity. All unit cells are less than 115 A.
>
> Thanks for any insight.
>
> Regards,
> Marilyn
>
> Marilyn D. Yoder
> Division of Cell Biology & Biophysics
> School of Biological Sciences
> University of Missouri-Kansas City
> 5007 Rockhill Rd.
> Kansas City, MO 64110-2499
> phone 816-235-1503
> fax 816-235-1503
>
