Recent replies to "BioGUID"

Reply from Egon Willighagen

Egon Willighagen — Thu, 02 Aug 2007 11:48:02 -0000

I found some time to play with RDF and set up a bioGUID-like service for molecules (InChIs to be precise):

http://cb.openmolecules.net/rdf/?InChI=1/CH4/h1H4

Roderic, what about:

inchi:InChI=1/CH4/h1H4

I’m sorry about the “InChI=” duplication, but that is an integral part of the InChI.

Reply from Egon Willighagen

Egon Willighagen — Sat, 07 Apr 2007 10:03:21 -0000

Roderic,

I want to play a bit with RDF in (bio)chemistry, after having read enough about the theory of things. DISCO seems to be a webpage(service) to do things. Do you know something GUI on one’s desktop instead, to work around those time out issues inherit to HTTP? Or does one directly have to go to tools like Jena?

Reply from Roderic Page

Roderic Page — Sun, 01 Apr 2007 17:29:08 -0000

There are links between entries, but these depend on the underlying sources. For example, the DOI http://bioguid.info/doi:10.1109/mis.2006.62 has no links because the CrossRef service I use doesn’t provide any. However, PubMed and GenBank records do have links. I good example is http://bioguid.info/gi:90296310 (the sequence DQ283141), which has links to the source organism, the publication, and the voucher specimen. These last two links (a DOI and a specimen id) are not in the original GenBank record, they are constructed. In the case of the DOI, I wrote a web service that parses the reference and looks for a DOI in CrossRef.

The metadata bioGUID extracts goes into a triple store. At some point it would be useful to expand the metadata returned. For example, one could add to the metadata for http://bioguid.info/doi:10.1206/0003-0090(2006)297%5B0001:TATOL%5D2.0.CO;2 the fact that it references the sequence DQ283141. I haven’t done this as the metadata would be incomplete unless I include all the sequences that paper refers to. I’m also trying to separate a GUID resolution service from a Semantic Web browser.

Disco is a nice tool. It’s 2 second timeout can case problems with bioGUID, because it often takes more than two seconds the first time a GUID is resolved (especially if more than one web service call is made). However, reloading the URI will usually get a result.

When I get the chance I will document bioGUID a bit more fully.