Re: [Treebase-devel] Rod Page on TreeBASEII

[William P. <wil...@ya...>]

On May 25, 2010, at 7:53 AM, Jon Auman wrote:

> We could :
> 1) keep the old site up and running and redirect all '/cgi-bin/foo' requests to the old server (seems backward)
> 2) Fix the 404 page to send users to the search page or custom 404 page (we should do this irregardless)
> 3) Create a lookup table that maps legacy taxonIDs to new ones. Seems like a one time thing, cause all new studies will not have legacy ID, right?
> 4) do #2 and not worry about the rest


The legacy APIs are described here:

http://treebase.peabody.yale.edu/treebase/urlapi.html

-- In the case of uBio, they used the "Link by Taxon Name" -- so that would be an easy rewrite. However, they have already said that they will be rewriting their outlinks to comply with our /phylows/. So it's hardly worth changing for them. 

-- In the case of NCBI, they used the "Link by Taxon ID Number" method, so the problem there is that we don't have a /phylows/find/taxon/ that uses TB1 legacy ids (which are stored in the tb1legacyid field in the taxonlabel table). So to make us compatible with NCBI, we'd need to write additional API code. Note that NCBI's outlinks are based on a very very old dump to them (e.g. back when TreeBASE had only 15,000 names, instead of the 100,000+ names that it has now). It would be vastly better if we get NCBI to change their outlinks over to /phylows/ and use NCBI taxids instead of legacy TB1 ids. Perhaps we should create a /phylows/ query that asks "give list of all NCBI taxids" (e.g. find all taxa with ncbi taxid > 0 that relate to at least one taxonlabel in TreeBASE), so that NCBI can dynamically update their outlinks rather than relying on periodic dumps.

bp