[Bioc-devel] Request for Comment: SNP Annotation package design

[Seth Falcon]

Hi Sean,

On 15 Feb 2006, sdavis2 at mail.nih.gov wrote:
> It seems like there is more information available than that in your
> proposed table structure.  Do you really want to drop those
> accessory data (like Avg Het. or from what database the snps are
> derived)?  

That's a good question.  What is in the proposal is the info
specifically requested by Rafael and his collaborators.

> If you stuck with the table structure from UCSC, then on
> the implementation side of things, one could envision being able to
> use RMySQL, RSQLite, or RdbiPgSQL to access the tables which could
> be easily obtained from bioc (RSQLite) or via a local database
> (which probably several of us already have running).

Interesting.  We are planning to use a database (SQLite) as the
backend for the SNP annotation.  If we are not going to integrate any
information from any other data source, then mirroring the UCSC
structure has a number of advantages.  

> I don't do much SNP work, so I shouldn't really be saying much, but
> disk space is pretty cheap, so it seems like more might be better
> (or at least not worse) here?

Yes, disk space is cheap, but network bandwidth is not.  There is some
value in keeping just the annotation data that is needed and having
smaller downloads, etc.

+ seth