Structure validation

1
Structure validation

• Everything that can go wrong, will go wrong.
• Everything that could go wrong has gone wrong.
• Especially with something as complicated as a
  protein structure.

2
What does WHAT_CHECK check?
Administrative errors. Crystal-specific
errors. NMR-specific errors. Really wrong
things. Improbable things. Things worth looking
at. Ad hoc things. www.cmbi.kun.nl/gv/pdbreport/ w
ww.cmbi.kun.nl/gv/servers/
3
How wrong is wrong?
4
Growth of the number of avoidable four sigmas
PDB FILES ERRORS
107
15K
3K
106
97 01 97 01
5
What did we find?
Trp with 90 degree angle between rings. His with
two N-s in side chain bound. Cys-Cys bridge with
one S. Asp and Leu conformation swapped. Lysine
neutral at pH 5.5. A molecule consisting of 628
waters. One molecule with four names. One name
used for four molecules. Thr with wrong C-beta
chirality. Missed cis-peptide bonds. Alternate
conformations are random.
6
What did we find?
B-factor refinement in TNT is random. His in
X-plor is always positive. Asp in CNS has protons
backwards. Planarity too relaxed in X-plor,
etc. Many waters placed at C-alpha. C-terminal
oxygen in middle of chain. Non-crystallographic
symmetry is real. Strand goes through core of
helix. Hexamer presented as monomer. Occupancies
above 1.0. Etc., Etc., Etc., Etc., Etc., Etc.,
Etc., Etc., Etc., Etc., Etc.,.
7
Close to 1.000.000 times

8
About 50 times

9
Your best check
10
Conclusions
Everything that could go wrong has gone
wrong. Errors are on a sliding scale. Error
detection can detect a lot, but surely not
everything (yet). Often we can detect that
something is rotten without knowing what is
rotten.
11
WHAT_CHECK by
Rob Hooft