dbFetch from EBI with bogus accession number returns valid record

Question

[Jeroen Van Goey]

I'm using dbFetch from EBI in a project. I construct URLs from the format

https://www.ebi.ac.uk/Tools/dbfetch/emblfetch?id=[ACCESSION_NUMBER]

to retrieve sequences, e.g.:

https://www.ebi.ac.uk/Tools/dbfetch/emblfetch?id=CO069431

Accession numbers need to be of the format 1 letter + 5 numerals OR 2 letters + 6 numerals,

so I added a test with the clearly bogus accession code XX.

To my surprise, the URL

https://www.ebi.ac.uk/Tools/dbfetch/emblfetch?id=XX

returned a valid response, namely the sequence for "Avocado sunblotch viroid, complete genome" with accession number J02020.

My questions:

1. Why does the accession number XX fetches the record with accession number J02020?
2. Is this a bug in the dbFetch software or incorrectly submitted data?
3. If so, should this be reported somewhere?

[shelly.deforte]

Interestingly, I couldn't find any other invalid code that returns something when it shouldn't - only 'XX'. Also, https://www.ebi.ac.uk/Tools/dbfetch/emblfetch?id=J02020 also returns the correct entry, so there are (at least) two pointers to this entry. Definitely looks like an error, but I'm not sure who the best person to report it to is.

[5heikki]

It's a rather old entry, which could mean that perhaps once XX really was its accession. Check for example this. Keep us updated..

..
CC   See also Symons R.H. Nucl. Acids Res. 9:6527-6537(1981) for
CC   sequence of an Australian isolate of ASBV (identical to
CC   unchanged sequence reported here). Accession number J02020; XX.
..