I'm struggling to find a fast way to access the metadata associated with a GRangesList. I can explicitly invoke:

lapply (GRangesList, mcols)

but that is very slow. Instead, I'm hoping to do something like

GRangesList$exon_rank

in order to get an IntegerList containing that data. (I then want to perform downstream operations on the list.)

More specifically, I want to extract a list of exon ranks from a TxDb object. The following code works:

refSeqDb = suppressWarnings (makeTranscriptDbFromUCSC (
"hg19",
tablename = "refGene"))

refseq2exons = exonsBy (refSeqDb, by = "tx")

refseq2exons = refseq2exons[, "exon_rank"]

exonRankList = lapply (lapply (refseq2exons, mcols), "[[", 1)

However, the final step--involving multiple calls to lapply--is extremely slow.

Hi Robert -- I think you want

ranks = unlist(refseq2exons)$exon_rank

which gives you the ranks, and 

ranksList = relist(ranks, refseq2exons)

which gives you an IntegerList of ranks grouped by transcript (get a plain-old-list as a third step, with as(ranksList, "list")). All will be fast.

The idea is that the GRangesList is represented by a single GRanges and a 'partitioning' vector that tells the GRangesList class how the elements of the GRanges are to be grouped to form the list. unlist()ing the GRangesList just removes the partitioning information (fast) and exposes the underlying GRanges with associated metadata. relist()ing takes the 'flesh' of the exon ranks and wraps them around the geometry of the 'skeleton' implied by the original GRangesList.

This

split (x = unlist (refseq2exons)$exon_rank, f = names (unlist (refseq2exons)))

is identical to as(ranksList[order(names(xx))], "list").

I figured out a somewhat non-intuitive way to do this using a split/unlist approach:

exonRankList = split (x = unlist (refseq2exons)$exon_rank, f = names (unlist (refseq2exons)))

This is less intuitive than I was hoping for, but it works and is reasonably fast.