Path: utzoo!attcan!utgpu!watserv1!watmath!uunet!wuarchive!cs.utexas.edu!tut.cis.ohio-state.edu!cica!iuvax!genbank!bio.embnet.se!mats
From: Mats.Sundvall@bio.embnet.se
Newsgroups: bionet.molbio.genbank
Subject: Re: Distributing GenBank over the Internet
Message-ID: <28.2583e1f4@bio.embnet.se>
Date: 11 Dec 89 16:20:52 GMT
References: <1989Dec7.213027.8591@phri.nyu.edu> <1989Dec9.190611.10877@ux1.cso.uiuc.edu>
Organization: Embnet node in Sweden at Biomedical Center, University of Uppsala, Sweden
Lines: 34

In article <1989Dec9.190611.10877@ux1.cso.uiuc.edu>, mike@pasteur.cvm.uiuc.edu (Michael Trogni) writes:
> Great idea, Roy, but I don't think most sites are network-wise
> enough to install NNTP and getting it running smoothly.  Aren't most
> of the biologists on these BIONET lists from BITNET?
> Also, does the data have a Expire time of never?
> 
> ps: new updates of genbank *are* available now from FTP in tar files.


Well, you could use the protocol in an installation that is not dependant
an news. I discussed this with Roy Omond from EMBL at a meating in Norway.
Maybe we try it out one of these days.

Maybe I shall clarify what I mean with not dependant on news.
If you write a program that works as an NNTP server but with another
TCP port, that easy to install, and that pick entries of the network
and put them into a database or wherever you want them, you could establish
a network for sequence distribution.
If you want to establish redistribution of sequences around the network
you also write a feed program.

You have to keep a list of NNTP ids that you have in your database. These ids
can be the accession number plus the revision date of the sequence.
Then you can implement ways of recieving updates of already existing sequences.
You can of course use the some hedaer to classify sequences if you are only
interested in parts of the database.
I also would like a checksum calculation in the protocol.

Not very hard to implement, but useful will it be in the light of Genbanks
"new" transaction protocol. Is this protocol better? Probably not as simple.

	Mats Sundvall
	University of Uppsala
	Sweden