Liking this very much! Thanx, Rajarshi! - Egon Willighagen

@Rajarshi a small bug: nothing was displayed with http://rest.rguha.net/usage... whereas the NCBI returns about ~9000 articles http://eutils.ncbi.nlm.nih.gov/entrez... - Pierre Lindenbaum

the site is a little slow, but a while ago I saw this tool for looking at PubMed trends on the internets: http://www.cotch.net/assed... - Michael Kuhn

@Pierre, thanks for the catch. I actually do get the correct data. The absence of bars seems to be a limitation of the Google Charts API and the max size of the chart image. If the X-axis is too long, the bars need to be thinner for it to fit into the plot area, and you can only go so thin and then loose readbility. Like I said - this is a quick hack :) Ideally I'd use matplotlib or R as the backend for plotting, but am lazy right now - Rajarshi Guha

interesting, and very cool. One question/comment, does the Google Charts API by default adjust the image so that the x-axis crosses at a non-zero point on the y-axis? I'm sure that breaks one of Tufte's rules about misleading graphs... - Andrew Su

@Andrew - yes. It took me a bit for me to realize that the count for the first year was not always zero! The Charts API does have a way to set the y range, but I'm going via the pygooglechart Python module which doesn't seem to support that. - Rajarshi Guha

i was about the link to my pubmed trends tool, but i see it has already been done :) it is indeed slow -- a clunky php proof of concept really. (it has been especially slow today because apache filled itself with processes that were trying to do i'm not sure what...) http://www.cotch.net/assed... - Joe Dunckley