[BioSQL-l] BioSQL at BOSC08 - Was Re: (no subject)

[James Procter]

Hilmar Lapp wrote:
>>> If you mean whether there were any questions about storing structure in
>>> BioSQL, no.
>> well - I was enquiring more generally - was there any discussion about
>> extending the BioSQL model for other kinds of bioinformatic objects ?
> 
> This was among the examples I showed for actual usage. In fact, I
> suspect that that's a fairly common usage pattern of BioSQL. Many of the
> papers citing BioSQL are by groups who have data persistence needs and
> work on sequences and annotation plus something custom they generate
> themselves. They use BioSQL as their sequence and annotation "module"
> that instantly solves that piece and how to get data in, and they add
> custom tables to model whatever types of data or results they have that
> are associated with that.
This is exactly the way I came to start using BioSQL (there are only so
many times the wheel should be re-invented, IMHO).

> One person wanted to accommodate pathway data (if I recall correctly?).
> I have so far been rather reluctant to add data types to the model that
> aren't really supported by at least one of the Bio* toolkits, as the
> main goal of BioSQL is to provide interoperable persistence for the Bio*
> toolkits rather than being a generic schema for all kinds of data. So I
> said I would look at it as soon as at least one toolkit has an object
> model for that kind of data.
> 
> I believe there are projects that aim to fill the latter need already
> and do so pretty well (Chado, as an example), and so rather than
> duplicating these efforts I thought that supporting Bio* persistence
> should take higher priority.
I have a rather similar opinion to yours, actually.

Part of BioSQL's attractiveness is its flexibility - which
may be reduced if additional tables are included since they will place
more constraints on the schema which in turn have a knock on effect on
the interoperation between BioSQL and the other Bio* object models. I've
recently been involved in a data-model based interoperability project,
and I can't say our experiences with trying to develop a unified
persistence model between heterogeneous datamodels were joyful.

I'd suggest that a wiki page is set up to describe any ad-hoc
'extensions' that BioSQL users think might be useful to the
community. If/When I get round to making any extensions myself then I'll
add them to that page, too.

Jim.