I do not remember the date or the PDB entry but it was during
the time that the PDB at BNL was including distance and angle
outliers in the checking report sent back to the depositor.
It was a not-too-large protein and there were perhaps half a
dozen outliers each on distances and angles which was typical of
an entry without 'problems'. So I sent the proposed entry to the
depositor and got a panicked call that something was wrong based
on the depositor looking at a display of the entry. By the next
day the depositor had figured out that s/he had decided to convert
to orthogonal for deposition and mistyped one of the cell dimensions
by 1 Angstrom. That length was about 135 so the error was less than
1% in one direction and I was very impressed that the depositor had
spotted it graphically.
After I did the appropriate linear algebra to correct the coordinates
I took a look at the distance and angle outliers. Of course they
were different but to my great surprise there were about the same
number of outliers for the 'bad' and the corrected entries. So based
on the checking at that time we could not tell the bad from the good.
I would be interested to know what would happen now with all the
additional checking that is available. Perhaps someone should do
that experiment.
Frances
=====================================================
**** Bernstein + Sons
* * Information Systems Consultants
**** 5 Brewster Lane, Bellport, NY 11713-2803
* * ***
**** * Frances C. Bernstein
* *** f...@bernstein-plus-sons.com
*** *
* *** 1-631-286-1339 FAX: 1-631-286-1999
=====================================================
On Thu, 16 Jul 2020, Gerard DVD Kleywegt wrote:
There was a case a few years ago (not too many though) where a 1.6 ?
structure had been solved using an incorrect value for the wavelength (~5%
too low, leading to a cell that was slightly too small for its contents to be
comfortable). It was later corrected so we could compare their validation
statistics. Some interesting observations:
- the geometry had been very tightly restrained so that didn't give a clue
about the cell error (WhatCheck only suggested a very small change)
- somewhat surprisingly (I thought) the Ramachandran plot did not improve in
the correct model (0.3% outliers in the wwPDB validation report), and the
sidechain rotamer outliers even got worse (from 1.5 to 2.5 %)
- the map looked surprisingly good for the incorrect cell
- however, RSR-Z told clearly that the map was not good enough for the
claimed
resolution - the model had 24% outliers! (3% in the corrected model which
still only put it at the ~50th percentile)
- another good indicator was the clashscore (went from 44 to 7)
- the original model did not include an Rfree, but the R-value (>0.3 at 1.6?
resolution) ought to have provided a clue to the crystallographers and
reviewers one would think
It would be interesting to see what would happen if the wavelength would be
set 5% too high.
--Gerard
On Thu, 16 Jul 2020, Clemens Vonrhein wrote:
Hi Robbie,
On Wed, Jul 15, 2020 at 07:23:15PM +0000, Robbie Joosten wrote:
At the same time if you have a a more relaxed approach to restraints
than you might find systematic deviations in bond lengths. A test
for that has been in WHAT_CHECK for decades and it actually works
surprisingly well to detect cell dimension problems.
Indeed.
That said, the problem is uncommon now.
Not so sure about that: we all rely on an accurate value of the
energy/wavelength from the instrument/beamline - and if that is off
(for whatever reasons) it will result in incorrect cell dimensions and
a systematic deviation from the various restraints.
This would even affect the best experiment done on the best crystal
... so fairly easy to spot at the refinement stage, especially if such
an energy/wavelength offset is constant over a long period of time on
a given instrument. To spot this at the data collection stage one
would hope that at some point a crystal with very pronounced ice-rings
will be looked at properly (and the fact these are not where we expect
them to should cause some head-scratching).
Cheers
Clemens
########################################################################
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing
list hosted by www.jiscmail.ac.uk, terms & conditions are available at
https://www.jiscmail.ac.uk/policyandsecurity/
Best wishes,
--Gerard
******************************************************************
Gerard J. Kleywegt
http://xray.bmc.uu.se/gerard mailto:ger...@xray.bmc.uu.se
******************************************************************
The opinions in this message are fictional. Any similarity
to actual opinions, living or dead, is purely coincidental.
******************************************************************
Little known gastromathematical curiosity: let "z" be the
radius and "a" the thickness of a pizza. Then the volume
of that pizza is equal to pi*z*z*a !
******************************************************************
########################################################################
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing
list hosted by www.jiscmail.ac.uk, terms & conditions are available at
https://www.jiscmail.ac.uk/policyandsecurity/
########################################################################
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list
hosted by www.jiscmail.ac.uk, terms & conditions are available at
https://www.jiscmail.ac.uk/policyandsecurity/
