Combine explicit peak sets into one common interval universe

Apply GenomicRanges::reduce() within declared groups, optionally require support from several input peak sets within each group, then take the union of retained group intervals. No count matrices are combined.

Usage

CombineATACPeaks(
  object,
  groups = NULL,
  min.support = 1L,
  min.width = 1L,
  max.width = Inf,
  genome = NULL
)

Arguments

object

A uniquely named list of GRanges. Each name identifies one peak-calling unit, not necessarily an independent biological replicate.

groups

Optional named character vector assigning every peak-set name to one support group, for example a condition. With NULL, all peak sets form one group. Support filtering is done within groups before their union.

min.support

Minimum number of distinct input peak sets supporting a reduced interval within a group. The default, one, does not require shared peaks across samples or conditions.

min.width, max.width

Inclusive width limits applied to the final union.

genome

Optional verified reference label. Known labels must agree; input sets without a reference label require this explicit declaration. Absent labels are not inferred from chromosome names. Known chromosome lengths must be compatible; the reduced union is checked again after it inherits known lengths from other inputs, before width filtering.

Value

A list with ranges in sorted order, named by canonical peak IDs, and a receipt recording input sets, support groups, filtering and dependency versions. Recount fragments over these ranges before comparing counts from previously different peak universes.

Details

Reduction uses ignore.strand = TRUE and min.gapwidth = 1L, so adjacent intervals are joined. Support counts distinct input sets contributing to a connected reduced interval; it does not mean every base has that support. Group-specific peaks can survive without support in other conditions. Final metadata records contributing sets and retained support groups, not caller p-values or a newly calculated consensus significance score.

See also

RecountATAC()

Examples

first <- GenomicRanges::GRanges("chr1", IRanges::IRanges(c(1, 201), width = 100))
second <- GenomicRanges::GRanges("chr1", IRanges::IRanges(c(51, 401), width = 100))
common <- CombineATACPeaks(object = list(first = first, second = second), genome = "toy")
common$ranges
#> GRanges object with 3 ranges and 2 metadata columns:
#>                seqnames    ranges strand |      input_sets  support_groups
#>                   <Rle> <IRanges>  <Rle> | <CharacterList> <CharacterList>
#>     chr1-1-150     chr1     1-150      * |    first,second             all
#>   chr1-201-300     chr1   201-300      * |           first             all
#>   chr1-401-500     chr1   401-500      * |          second             all
#>   -------
#>   seqinfo: 1 sequence from toy genome; no seqlengths
common$receipt$input_sets
#>     name group n_peaks
#> 1  first   all       2
#> 2 second   all       2