Hi :

I have GRangesList grs to find overlap and other meta-analyses, after several workflows, I obtained resGR, which needed to be relisted. However, I come up with the solution to relist `resGR` as GRangesList, but I got an error. My solution works fine with data.frame, but can't work with the S4 class. Can anyone point me out how to fix this error? Is there any simplest way to relist GRanges respect to GRangesList?

Here is input GRangesList :

grs <- GRangesList(
    bar= GRanges(seqnames = Rle("chr1",4),
                 IRanges(c(8,18,33,53), c(14,21,39,61)),
                 score=c(48,7,10,8)),
    cat = GRanges(seqnames = Rle("chr1",4),
                  IRanges( c(6,15,20,44), c(10,17,34,51)),
                  score=c(54,21,14,12)),
    foo= GRanges(seqnames = Rle("chr1",2),
                 IRanges(c(11,43), c(36,49)), score=c(49,13))
)

After several workflows, I obtained resGR:

resGR <- GRanges(seqnames = Rle("chr1",9),
                 IRanges(c(53,8,6,15,11,44,43,44,43), 
                         c(61,14,10,17,36,51,49,51,49)),
                 score=c(8,48,54,21,49,12,13,12,13))

I want to relist resGR respect to grs, so I come up with this solution (my attempt might not be desired, perhaps there is  might have a more straightforward way to do this); Here is my solution:

lapply(grs, function(x) resGR[do.call(paste, resGR) %in% do.call(paste, x),])

I have this error:

Error in do.call(paste, resGR) : second argument must be a list

Perhaps my attempt is not feasible, but it works well if use data.frame. Is there any simple and vectorized approach to get this done?  Any idea? Thanks a lot.

Best regards:

Jurat

How about this?

​matches <- lapply(grs, match, resGR)

reList <- lapply(matches, function(x, res) {
  x <- na.omit(x)
  res[x]
}, resGR)

reList <- as(reList, "GRangesList")

The first line takes the input GRangesList, and for each entry finds any matches in your output GRanges object.

The second part then takes the resulting list of matches, and gets the appropriate entries from resGR and returns it as a normal list of GRanges objects.  

Finally we convert that to a true GRangesList

The output I get is:

>
GRangesList object of length 3:
$bar 
GRanges object with 2 ranges and 1 metadata column:
      seqnames    ranges strand |     score
         <Rle> <IRanges>  <Rle> | <numeric>
  [1]     chr1  [ 8, 14]      * |        48
  [2]     chr1  [53, 61]      * |         8

$cat 
GRanges object with 3 ranges and 1 metadata column:
      seqnames   ranges strand | score
  [1]     chr1 [ 6, 10]      * |    54
  [2]     chr1 [15, 17]      * |    21
  [3]     chr1 [44, 51]      * |    12

$foo 
GRanges object with 2 ranges and 1 metadata column:
      seqnames   ranges strand | score
  [1]     chr1 [11, 36]      * |    49
  [2]     chr1 [43, 49]      * |    13

-------
seqinfo: 1 sequence from an unspecified genome; no seqlengths

One obvious caveat is that this throws away anything in your resGR object that can't be found in the start data, but it's not clear to me what you'd want to do with it anyway.