[Biopython-dev] KEGG support

[Renato Alves]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

- From Giovanni Marco Dall'Olio on 02/10/2010 10:13 PM:
> Hi,
> I had a terrible experience with parsing Kegg pathway's files: in the
> end I discovered that the files that are stored in their ftp don't
> correspond exactly to the diagrams that you can find in the web
> interface, as for example biochemical interactions don't have
> directionality while if you look at them on kegg/pathway you will see
> arrows.

I haven't used pathway files yet so I'll be careful when I reach them :)
Have you mentioned this aspect to the KEGG maintainers?

> Some time ago I proposed to implement something similar to what you have
> said for kegg/pathway, but in the end I abandoned the effort, because I
> had problem both with suds and SOAPpy, and I wasn't satisfied by the
> annotations in KEGG.
>  
> If you are serious about that I may help you, but I can only work on the
> weekends and you should tell me exactly what I have to do :-) 

Hehe, I can only tell you once I get my hands dirty. I'll keep my code
on github to maximize interaction. I'll get back at you when I get the
first working draft for GENES. Thanks for the hand ;)

> Are you sure? I tried it on KEGG an year ago and I was having problems
> to execute slightly more complex queries. If you look at suds's bug
> tracker, you will find some reports by me, like this one:
> - https://fedorahosted.org/suds/ticket/213

As of suds revision 658 I can no longer reproduce the error in the ticket.

> I remember that I was looping between the KEGG support centre and the
> suds bug tracker; both were very responsive to feedback and very keen to
> answer me, but in the end they didn't speak to each other and the bug
> reports that I have filed are still unfixed.
> 
> Which library can you use for the soap queries? I had the feeling that
> SOAPpy (which I think it is included in the standard lib) worked well
> with KEGG, however it development has stopped many years ago
> (http://sourceforge.net/projects/pywebsvcs/files/SOAP.py/), it is a mess
> if you want to use it behind an http_proxy (I should have a patch
> somewhere if you are interested) and I am sure it won't be kept
> compatible with the future versions of python.

SOAPpy doesn't seem to be in the standard lib, at least I don't have it
out of the box here. Only as external package in the repository.

> Another alternative may be beautiful soup, but I have never tried it.

I've only used beautiful soup as HTML cleaner/formatter, like HTML tidy.
I wasn't aware that it could be used for SOAP stuff. Are you sure about
this?

> This question on stackoverflow may provide you some ideas:
> http://stackoverflow.com/questions/206154/whats-the-best-soap-client-library-for-python-and-where-is-the-documentation-fo
> 
> I am not sure about which is the standard soap library for python, and
> which one is included in the standard lib. If you are going to use
> SOAPpy, it is a bad bet toward compatibility and maintenance for the
> future releases. Suds is the best option but it is not in the standard
> lib, and they still have to fix the bugs I have reported an year ago. I
> have the feeling that there is no good alternative for python.

I'll wait for your opinions. I don't want to sound religious about suds. :P

> Moreover, the WSDL functions that I have seen for KEGG are not
> especially useful. They seems to allow for the basic queries, but for
> most of the tasks it is better to download the ftp locally and work there.

Well if you just want a quick check on something the API still gives
better/quicker results than downloading the stuff via FTP. Given the
size, probably the load of the server and the fact that I'm on the other
side of the globe, I got an ETA of close to 20 hours when downloading
the genes.tar.gz file which is only a few GB in size.

Renato
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.14 (GNU/Linux)

iEYEARECAAYFAktzWgoACgkQYh11EUYTX9Rp6QCfaHf6Ic3uT/npDw2o8l9F+8Kk
RtgAnjNXGxcrfvh48dcdFf6G4wK9+PNI
=vpUY
-----END PGP SIGNATURE-----