Hi Kim,

On Fri, Aug 17, 2007 at 03:04:27PM +0100, Kim Henrick wrote:
> The discussion on archiving image data sets -
>  I guess that less than 1% of the image sets for PDB entries
>    are useful to software development (and can be got privately)

I would think much much more would be useful for software development:
the 1% will maybe be the real difficult ones or the extremely
high-resolution ones - which is interesting for development as well.

But a large extent of software development goes into methods for the
'everyday' kind of structure ... the 99%. It is fairly easy for us
getting hold of very difficult datasets and extremely good datasets:
but try and find a boring, standard dataset. However, these are the
ones that new methods need to improve upon as well. Otherwise we end
up with methods and software that work for the 5 special cases a year
and not all the others.

We routinely use ALL the PDB entries (where structure factors are
available) for running various tests and analysis (and I know other
software development groups do this on a regular basis as well).

If there was a mechanism for depositing raw images it might start in
the same way as the deposition of structure factors: fairly small
scale and only done by a few. It might not even have to start as a
requirement (structure factors weren't a requirement either at the
beginning afaik). But just having the possibility might open up new
insight: not only into the particular structures or software
development, but also into how to handel this data.

And yes, I agree: a distributed system (a la doi) would be much, much
better than a central system. At least the existing synchrotron or
SG-center infrastructure could be re-used.

Cheers

Clemens

-- 

***************************************************************
* Clemens Vonrhein, Ph.D.     vonrhein AT GlobalPhasing DOT com
*
*  Global Phasing Ltd.
*  Sheraton House, Castle Park 
*  Cambridge CB3 0AX, UK
*--------------------------------------------------------------
* BUSTER Development Group      (http://www.globalphasing.com)
***************************************************************

Reply via email to