[Bioperl-l] genbank mirror

[Carlo Lapid]

Hi,

I'm trying to set up a local mirror of a large part of the Genbank database.
For users to access the local database, I need to create a web-based search
tool, much like Entrez of NCBI, or SRS of EBI; that can parse the Genbank
flat files I've downloaded based on a query entered by the user.

I'm trying to use Bioperl to create this from scratch, but I'm having a very
hard time, especially since I want the user to have reasonable flexibility
in customizing his search. The best that I've been able to accomplish is a
search function that retrieves genbank sequence objects based on their
primary IDs or accession numbers; by using the fetch method of the
Bio::Index::GenBank module. But this doesn't help users who don't know the
exact IDs for the sequences they want.

Can anybody suggest a way to use Bioperl to search for an ordinary word or
phrase, like "16S gene", which could be matched against the description
field, or the entire genbank entry? (Alternatively, is there some other
freely available tool or software that can do this?) I've been scouring the
Bioperl documentation, but I couldn't find anything. I just need to be
pointed in the right direction. What I thought was a relatively simple
problem has been driving me crazy for days; if anybody has any suggestions I
would really, really appreciate it.