[Bioc-devel] eSet class and two color arrays

[Seth Falcon]

Wolfgang Huber <huber at ebi.ac.uk> writes:
>>    So would think you'd want an assayData slot containing four
>>    matrices:
>>        
>>        Rb  Red background
>>        Rf  Red foreground
>>        Gb  Green background
>>        Gf  Green foreground
>> 
>>    Each matrix has dim genes (rows) by samples (coloumns).  A column
>>    in any of these matrices refers to the same biological sample and
>>    corresponds to a single row in phenoData.
>
> No, different samples are on the red and green channels (so these four
> matrices correspond to two phenoData object rows).

Thanks.  That along with some offline conversations give me a much
clearer picture of what is needed here.

> One way to close this thread is Robert's answer, and I do not
> object.
>
> (Still it sticks that "raw" data is not always that different from
> "processed" data, whereever you draw the line, and reusing what is there
> as much as possible could, in an ideal world, where people care, be
> better than making up, maintaining and making people learn completely
> different data structures. 

I don't disagree that reusing code is a good goal, but as Robert
explained, the eSet class is intended for processed, not raw, data.

I think a good approach would be for an interested party to consider
building a rawSet class that made use of some of the same component
objects as eSet (AnnotatedDataFrame and AssayData).  I'm pretty sure
one does not want a subclass of eSet because the behaviors one will
want, even in terms of subsetting, I think are quite different.

+ seth