[BioSQL-l] RE: BioSQL and Bioperl-db inching towards release

[Marc Colosimo]

On Thu, 29 May 2003, Hilmar Lapp wrote:

> (Technically, this is more along a snow-flake than a pure star-schema
> design.)
> 
> You don't want to use this for natively hosting your raw values and LIMS
> annotation. Instead, you have another 'real' db for expression data etc
> and once a week or so you repopulate the above table.
> 
> You could use the same table and the same principle for other biological
> data your lab(s) produce(s) so long as they are numeric.
> 
> As for a real native expression database in Biosql, I am happy to see
> someone else drive this; as things stand now I won't. I maintain the
> points I've expressed earlier though, namely let's not re-invent the
> wheel here, and if you do, that wheel in biosql should be generic and
> technology agnostic (which is not without significant challenges).
> 
> I'm not sure but I believe I failed to list a couple of links to open
> source gene expression databases before when you asked for that. Here
> goes, but quite frankly I'm surprised that you couldn't find them on the
> web (they are all googleable).
> 
> RAD (RNA Abundance Database) is part of GUS from UC Penn:
> http://www.gusdb.org/, RAD is a hyperlink there.
> 
> GeneX (also check out GeneX-lite linked from the page below)
> http://genex.ncgr.org/
> 
> TIGR's entire Microarray analysis suite, includes database and lots of
> beautiful software that supposedly even works (as they use it
> themselves)
> http://www.tigr.org/software/tm4/
> 
> BASE (BioArray Software Environment) (there are local installations of
> BASE running at various institutions world-wide already)
> http://base.thep.lu.se/
> 
> I'm sure none of the above is perfect under every condition, so you may
> still reach the conclusion that you want to roll your own. I just
> wouldn't without having looked closely at them.
> 

I did find most of those (plus others). I didn't find BASE which looks 
interesting (opensource, but written in C++, ack). TIGRs also is 
promising. However, neither fully support affymetrix data. Someone just 
tried it for BASE and posted a message about their failure and I haven't 
seen anything concrete for TM4 (there is a tool for GenePix to TIGR 
ArrayView). I think one of the big problems is the large difference 
between DNA/oligo chips and Affy's Oligo chips (PM/MM stuff). Some papers 
have suggested to just forget about the MM and use the PM for analysis. 
But that still leaves 11+ spots per gene. I think I basically have to roll 
my own code to get things to an expression level stage (one number per 
gene). BASE looks like a good place to get a start. I don't trust myself 
writting any stats and BASE can use R (therefore, bioconductor).

Marc