Re: Time for CA-only models to be abolished ? (was: Re: [ccp4bb]:Howto deal with sidechainatoms with low electron density?)

[Bart Hazes <bhazes@ualberta.ca>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

On Tue, 26 Nov 2002, Gerard DVD Kleywegt wrote:

> ***  For details on how to be removed from this list visit the  ***
> ***          CCP4 home page http://www.ccp4.ac.uk         ***
>
>
> that's all nice and dandy, but:
>
> (a) what good is information that is unreliable and unverifiable ? is a
> CA-only model with tracing errors really of any use ? is it (biological)
> signal or is it (biological) noise ? i tell students to compare 1PTE and 3PTE.
> then i ask them to have a look at 2PTE ... how much research effort and money
> has gone wasted due to incorrect models we'll probably never know.

Yes of course there are very valuable CA models where a researcher truly
couldn't determine the full atomic model. Some examples were mentioned by Ed
earlier. Basically if you can't determine chain direction then CA-only is the
best you can do. Clearly information content will be far less than a full atom
structure but far more than many of the theoretical models, low resolution EM
blobs (no disrespect intended), or schematic diagrams showing six bars to
indicate six transmembrane helices. Whether the new knowledge warrants
publication depends on the area of research. As always the important thing is
to limit the detail in the interpretation to the detail provided by the
experimental data. But if doing so gives significant structural insight then I
have no problem in publishing such work.

> (b) there are cases of CA-only models with obviously refined temperature
> factors. this suggests that people have 'grep'-ed the CA lines out of their
> PDB file and hence intentionally obfuscated their results.

This should be the real issue. People should not be allowed to hide their
results by depositing only CA coordinates. The easiest way to enforce this is
to demand deposition of Fobs and experimental&model phases. Honest CA
depositers wont mind because they have nothing to hide (apart from the fact
that they give potential competitors a better molecular replacement model).

> for a very recent example to which both (a) and (b) apply (plus absence of
> structure factors of course), see FEBS Letters, vol 525, pp 174-175 (2002)
> [plus the reply by the crystallographers on pp 176-178 which, if it wasn't so
> serious a matter, would be laughable]

There will always be exceptions to the rule but on the whole crystallography
is doing a great job in elucidating protein structure. Depositing structure
factors should prevent intentional misuse and correct accidental mistakes. At
the same time it should solve the CA-only problem without penalizing the valid
uses of CA-only models.

Bart

===============================================================================

Assistant Professor
Dept. of Medical Microbiology & Immunology
University of Alberta
1-15 Medical Sciences Building
Edmonton, Alberta, T6G 2H7, Canada
phone:	1-780-492-0042
fax:	1-780-492-7521

===============================================================================