Skip to content
Prev 3603 / 21307 Next

[Bioc-devel] SummarizedExperiments

I'll respond here, though I've read the rest of the thread.
On 09/06/2012 02:26 PM, Kasper Daniel Hansen wrote:
I shied away from these intentionally, using colnames, rownames, colData 
respectively. These emphasize the matrix-like behavior of a 
SummarizedExerpiment. Also the 'features' of featureNames has a 
technical meaning in the MAGE model of microarrays (and is, 
embarrassingly, used incorrectly in ExpressionSet). For newcomers it is 
also not transparent what the 'p' in 'pData' stands for, and for the 
more establishing user it will come as a surprise that the nicer design 
of DataFrame means that phenoData is already implied by colData -- look 
in mcols(colData(.)) for the equivalent of varMetadata.
Tim mentioned that seqlengths and seqlevels are implemented already. The 
remainder are all operations on GRanges or GRangesList, and I think it's 
reasonable to implemented them (and others in the GRanges / GRangesList 
API). I want to give some thought to doing this efficiently; thanks for 
the pointer to hasGRanges.
Tim mentions sub-setting and subsetByOverlaps. Conceptually I like the 
idea of sub-setting one GRanges by another. I'll definitely implement 
subsetByOverlaps, which I think of as approximately se[countOverlaps(se, 
gr) != 0,], and will also rethink [ sub-setting.

Thanks for the input.

Martin