[Pdbwiki-devel] Uniprot mapping & error reports question

Dan Bolser dan.bolser at gmail.com
Tue Jul 28 07:03:26 EDT 2009


2009/7/28 Henning Stehr <stehr at molgen.mpg.de>:
> Hi Dan,
>
> Jose and I went through the Uniprot mappings.
>
> In fact, out of 83455 polymeric chains (which could potentially
> contain mappings to Uniprot),
> 60316 are correct,
> 14158 have no Uniprot cross reference in the CIF file
> 8981 have a broken link, because there was a Uniprot cross reference,
> but the value was NULL, which we didn't check for (see e.g. 2cad)
> 1 had garbage in the Uniprot reference field in the cif file, so we
> wrote a comment in the Wiki about it (3g6d H)
>
> So the broken links do exist in about 11% of the chains (or ~12% of
> the entries). We have fixed the script such that this should not
> happen for future entries. If time permits, Jose

Wow! Good work!

It's so great to get people to critically look at the site. The more
eyes the better! We should set up some kind of 'reviewer commons' so
that the effort put into the scientific review process gets due
credit.


> will also update the old entries with the search-and-replace-script.

Is it tough to create an 'update this list of files' script? I know we
have an 'update script', which is based on modification time stamps,
but a general 'recreate this given bunch of entries' script could be
useful in such cases.


> I have another general question:
>
> Shall we compile a list with "errors collected by the PDBWiki
> community" and send it to the PDB?

Great idea. This could be very useful. In the reply, I think it would
be better to later say "We did send an error report ...".


> At the same time we could ask the PDB to link to our site.

That would be great. The more links (to or from us) the better.


> Anyways, this is not the highest priority, so we should probably
> finish the other stuff first.

Assuming it won't take too long, I don't see why we should delay doing
that. Getting a fresh round of comments back from PDB-L would be
really great.

However, I think we need to guide users from the homepage to the error
categories more clearly (this could take some time to re-design).
Currently the error categories are pretty well hidden from the
homepage. Since its a major site 'feature' I think we need at least a
couple more big links:

* Browse comments (or errors)
* Add comments (or errors)


Also, I've noticed a few FAQ's on the PDB-L, but I just don't like to
bother people too much with the boring "this question got asked
already..." reply.


Thanks again for checking the Uniprot mappings,

Dan.

> Henning
>
> _______________________________________________
> Pdbwiki-devel mailing list
> Pdbwiki-devel at bioinformatics.org
> http://www.bioinformatics.org/mailman/listinfo/pdbwiki-devel
>




More information about the Pdbwiki-devel mailing list