I know how to search for a pubmed identifier in citeulike. Since yesterday I have tried to do the opposite that is to get all pubmed identifiers of collection in CiteULike. I know use Hubmed, to get the pmid's but this is quite laborative. Since the information seems to be captured in CiteULike I am wondering if you could get them with a single api call.
Just grab the collection as JSON: /json/user/fred Pretty much any modern programming language can deal with that, e.g., Python.
(The CiteULike forum is a more reliable place to ask such questions)
Fuller details are on the help pages http://wiki.citeulike.org/index.php/Importing_and_Exporting#JSON
For performance reasons, you can only grab a max of 250 records per call (default 50), so you'll need to "page" through until you get fewer than the expected number returned.