[Bioperl-l] Microarray ANOVA module

[Alvis Brazma]

> 
> 
> What I would love to see is some discussion of the "business objects" and 
> interfaces - by business objects I mean objects which are indepdent of 
> whether they were built by files/databases/xml/whatever, for example, one 

I think that MAGE-OM is such an object model, which is completely 
independent of databases or any particular implementation (see 
www.mged.org/mage)


> might have:
> 
>   (these are all made up things)
> 
>   # assumming these are all interface definitions
> 
>   Bio::Array::PhysicalDescriptionI    # which points go to which elements
>      ::Array::ElementI                # what is actually on the array
>      ::Array::ExperimentDescriptionI  # description of an experiment
>      ::Array::CapturedExperimentI     # A set of data, with a 
>                                       # PhysicalDescriptionI and a 
>                                       # ExperimentDescriptionI, and of 
>                                       # course the data values!
> 
> 
> 
> Of course, ArrayExpress and friends are MIAMI compliant and MAGE-ML based,
> both of which are clearl things to support, but I strongly believe that
> the full MAGE-ML model should not be exposed as just "400 objects you
> might want to use for describing array experiments" but instead mapped to
> some clear cut interfaces and hide the complexity sensibly (eg, one might
> have)
> 
> 
>    Bio::Array::MAGEML::DataValue
>                      ::TemperatureValue
> 
> 
> etc
> 
> 
> or indeed, have this outside of the Bio:: namespace as MAGEML may 
> auto-generate the objects and one just needs to write bridging cases to 
> the interfaces that we should agree on.
> 
> 
I'm not 100% sure if I uderstand your point... How would you autogenerate
these 400 or so objects from less granular description? MAGEstk project is 
working to generate Perl, Java and C++ objects from MAGE-OM, and in fact 
the Perl version already works and we are using it for MAIMExpress. I can 
understand that for human use the model should be conceptualised, to 
extent we had try to do it with 
http://www.mged.org/Workgroups/MAGE/introduction.html and a publication is 
comming out on this, but I don't see how a computer can autogenerate these 
objects from less granular description? Or I guess, this is probably 
something else you are suggestiong?

In any case, we have a MAGE mailing list to which you can subscribe 
from www.mged.org/mage, and these discussions would be useful for the 
list. You have to be a member to send an e-mail to the list. 

Cheers, 
- Alvis



> 
> I am not a pro here, so I'd be interested in anyone's thoughts, but of 
> course, we need to discuss how this should work inside bioperl as much as 
> possible/sensible so groups have a common framework/concept set to work 
> from
> 
> 
> 
> 
> 
> 
> 
> > 
> > It allows data loading and retrieval from/to mysql. At this time it knows of
> > Affimetrix data format for loading, and reports to files able to be analyzed 
> > by GeneSpring.
> > 
> > Jaim
> > -- 
> >  Dr Jaime Prilusky                | Jaime.Prilusky@weizmann.ac.il
> >  Weizmann Institute of Science    | fax: 972-8-9344113
> >  76100 Rehovot - Israel           | tel: 972-8-9344959
> > 
> >  info URL http://bioinformatics.weizmann.ac.il/jaime_prilusky.html
> >  OCA is at http://bioinformatics.weizmann.ac.il:8500
> > _______________________________________________
> > Bioperl-l mailing list
> > Bioperl-l@bioperl.org
> > http://bioperl.org/mailman/listinfo/bioperl-l
> > 
> 
> -----------------------------------------------------------------
> Ewan Birney. Mobile: +44 (0)7970 151230, Work: +44 1223 494420
> <birney@ebi.ac.uk>. 
> -----------------------------------------------------------------
> 
>