reduceByRanges summarize behavior with vector-returning MAP

[vjcitn]

Suppose the MAP returns a 2-vector and summarise=TRUE

The resulting assay looks like

assay(rr2)
[,1] [,2] [,3]
ENSG00000111424 Numeric,2 Numeric,2 Numeric,2
ENSG00000172216 Numeric,2 Numeric,2 Numeric,2
ENSG00000124731 Numeric,2 Numeric,2 Numeric,2

which is quite useful ... the [i,j] element is a 2-vector. But it might be more
useful to have a pair of flat matrices in the assay slot. Do you think that would
take a lot of work?

[mtmorgan]

Can you provide a reproducible example, maybe building off example(reduceByRange)?

[vjcitn]

let

MAP2 = function (range, file, ...) 
{
    requireNamespace("GenomicFiles", quietly = TRUE)
    param = Rsamtools::ScanBamParam(which = range)
    rep(sum(runValue(GenomicFiles::coverage(file, param = param)[range][[1]])), 
        2)
}

then

rr2 = reduceByRange(grl, fls, MAP2, summarize=TRUE)

> assay(rr2)
          ERR127306 ERR127307 ERR127308 ERR127309 ERR127302 ERR127303 ERR127304
ERR127306 Integer,2 Integer,2 Integer,2 Integer,2 Integer,2 Integer,2 Integer,2
ERR127306 Integer,2 Integer,2 Integer,2 Integer,2 Integer,2 Integer,2 Integer,2
          ERR127305
ERR127306 Integer,2
ERR127306 Integer,2

This just illustrates the situation. The current behavior is useful, but, for example,
tweaking the example just a little bit,

> reduceByRange(grl, fls, MAP, summarize=TRUE) -> rr3
> assay(rr3)
          ERR127306 ERR127307 ERR127308 ERR127309 ERR127302 ERR127303 ERR127304
ERR127306 ?         ?         ?         ?         ?         ?         ?        
ERR127306 ?         ?         ?         ?         ?         ?         ?        
          ERR127305
ERR127306 ?        
ERR127306 ?