Dear All,

To add my voice to those who wrote crystallization artifacts happen - we 
just witnessed one today. A postdoc from another lab tried crystallizing 
a protein for months. Today, during data collection from his poorly 
diffracting crystals, one of our guys (Dr. Porebski) scaled the data and 
checked if there is a structure with the same Sp Gr and unit cell 
parameters in the PDB. And there was one - 4ZNZ (Escherichia coli 
carbonic anhydrase (YadF) in complex with Zn - an artifact of 
purification), deposited by our group two years ago. Quick molecular 
replacement showed that it is indeed the protein we had in the beam today.

The protein looked good on the SDS gel. The band for YadF was definitely 
less than 1% of the protein sample (less than one percent!!! if the gel 
would not be overloaded, nobody would see the band), but it still 
crystallized. We speculate that YadF from Escherichia coli was 
co-purified with the target protein due to relatively strong 
protein-protein interactions despite multiple purification steps.

We did not use ContaMiner, though. Instead, we just used the search of 
the unit cell parameters in the PDB, which is much faster (but works 
only if the structure of this particular artifact in the same SpGr is in 
PDB! otherwise, one should use ContaMiner or similar service). This 
feature is implemented in HKL3000, but it can also be done quickly on 
the PDB webpage. The procedure, as well as cases with other 
crystallization artifacts, is described in:

Protein purification and crystallization artifacts: The tale usually not 
told. (2016) Protein Sci. 25: 720-733

Today's example clearly shows that depositing these artifacts can 
greatly help others (it took just a few minutes to realize we had an 
artifact). Therefore, I would like to encourage everyone to deposit 
their artifacts to the PDB, especially if these artifacts were 
crystallized with unit cell parameters not reported previously. An 
increase in the size of the library of crystallization artifact 
structures deposited to the PDB can make troubleshooting of new 
artifacts much easier, and save much effort for those who are new (or 
not very new!) to such cases.


With best regards,
Ivan Shabalin, Ph.D.
Research Scientist,
Department of Molecular Physiology and Biological Physics,
University of Virginia,
1340 Jefferson Park Avenue, Pinn Hall,Room 4223,
Charlottesville, VA 22908

On 11/23/2017 02:35 PM, [log in to unmask] wrote:
> Dear Stefan,
> 
> Just a couple of thoughts:
> 
> - first of all I think that Gerard is absolutely right, it would have been
> nice to raise such issues first with the developers. In my experience,
> Staraniso does a fantastic job if used correctly.
> 
> - but if you're OK with public trials, may I ask: why on Earth would anybody
> need ContaMiner? Are you trying to offer some sort of computational cure for
> sloppy biochemistry? There is zero point in crystallizing crap samples, sorry
> to say this. In my 17 or so years in Strubi I've never heard of anybody
> crystallizing a "contaminant", being it a purification tag or whatever.
> 
> I suppose this might have happened to somebody you know, hence the motivation
> to spend time on the bizarre ContaMiner. Which is a pity, a silly outcome
> would only teach people to do their job (or train their robots) properly.
> 
> Best wishes,
> 
> Radu
> 

----