About two weeks ago, tipped off by Neil, I heard about Google App Engine. I managed to get a beta account, and I’ve finally had a chance to do something (hopefully) useful with it.
In the absence of any quickly achievable ideas for a bioinformatics app, I ported over the OpenRef application I wrote on top of TurboGears a few months back.
or when there is no volume number, like:
will query PubMed and forward you to the DOI. Usually this means you will be forwarded to the electronic version of the article at the publishers site. The nice thing about these OpenRef style URLs is that unlike the DOI (eg doi:10.1093/molbev/msm031) or PubMed ID (eg PMID:17329230), they are easily guessable if you have the biblographic reference in front of you. See Noel’s original OpenRef post and the “About” box at ResolveRef for more details.
Here’s a reference and the corresponding ResolveRef URL you can click on to test it out:
(wow … I really should have chosen one without such a large forest of authors )
I’ve called it ResolveRef rather than OpenRef, since based on the discussion on Noel’s blog, it became clear that this method of mapping PubMed queries to URLs doesn’t completely fulfill Noel’s vision of what an OpenRef resolver is meant to be. It looks like this discussion is on-going (and has been had before) … in some ways it looks like this style of reference location by URL has missed the boat since DOIs and the CrossRef database are already well established and prevalent.
As far as how difficult the porting from Turbogears was … well, for an app like this with no database access, and essentially no web interface templates (I only added the front page with a form for the ResolveRef version, the original app served no actual html pages) … it was pretty straightforward, and I nice way to learn some of the ins-and-outs of App Engine. The biggest trick was hacking the Biopython Bio.EUtils.ThinClient class to use Google App Engines urlfetch rather than urllib2, which amounted to changing something like:
import urllib2 return urllib2.open(url)
from google.appengine.api import urlfetch fetched = urlfetch.fetch(url) return cStringIO.StringIO(fetched.content)
Bugs and feature requests are welcome … but I don’t make any promises about fixing/adding them. I’ll release the code soon, when I get some time to clean it up for public consumption