n x p

[ellispatrick]

I am writing this issue to publicly express my support for a future transition to a n x p representation of data in bioconductor. I appreciate that this will likely never be realised through SummarizedExperiment, but regardless,

p x n historically makes sense, however, p x n creates an unnecessary conceptual hurdle for new users of bioconductor and developors outside of the bioconductor community.

[vjcitn]

Thanks Ellis. Can more details be provided about the conceptual hurdles? X[G,S] is an arbitrary construct that has been at the base of the ecosystem for 2 decades: X is an integrative representation of assay outcomes and sample characteristics, G is a predicate on p genomic features, S on n samples, the bracket operation is endomorphic. The "back end" can take many different forms and is fairly often "not" a p x n array. If we are deliberate in use of accessors the p/n concepts are hidden.