Merge pull request #97 from snystrom/#96

Implements strandedness for filter_by_overlap() family. Closes #96.

NAMESPACE

@@ -49,9 +49,11 @@ S3method(filter,GroupedIntegerRanges)
 S3method(filter,Ranges)
 S3method(filter_by_non_overlaps,GenomicRanges)
 S3method(filter_by_non_overlaps,IntegerRanges)
+S3method(filter_by_non_overlaps_directed,GenomicRanges)
 S3method(filter_by_overlaps,DeferredGenomicRanges)
 S3method(filter_by_overlaps,GenomicRanges)
 S3method(filter_by_overlaps,IntegerRanges)
+S3method(filter_by_overlaps_directed,GenomicRanges)
 S3method(find_overlaps,GenomicRanges)
 S3method(find_overlaps,IntegerRanges)
 S3method(find_overlaps_directed,GenomicRanges)
@@ -210,7 +212,9 @@ export(disjoin_ranges_directed)
 export(expand_ranges)
 export(filter)
 export(filter_by_non_overlaps)
+export(filter_by_non_overlaps_directed)
 export(filter_by_overlaps)
+export(filter_by_overlaps_directed)
 export(find_overlaps)
 export(find_overlaps_directed)
 export(find_overlaps_within)

R/ranges-overlap-filter.R

@@ -11,7 +11,7 @@
 #'
 #' @details By default, `filter_by_overlaps` and
 #' `filter_by_non_overlaps` ignore strandedness for [GRanges()]
-#' objects. The argument `maxgap` is the maximum number of positions
+#' objects. To perform stranded operations use `filter_by_overlaps_directed` and `filter_by_non_overlaps_directed`. The argument `maxgap` is the maximum number of positions
 #' between two ranges for them to be considered overlapping. Here the default
 #' is set to be -1 as that is the the gap between two ranges that
 #' has its start or end strictly inside the other. The argugment
@@ -39,8 +39,12 @@
 #'
 #' filter_by_overlaps(query, subject)
 #'
+#' filter_by_overlaps_directed(query, subject)
+#'
 #' filter_by_non_overlaps(query, subject)
 #'
+#' filter_by_non_overlaps_directed(query, subject)
+#'
 #' @rdname ranges-filter-overlaps
 filter_by_overlaps <- function(x,y, maxgap = -1L, minoverlap = 0L) {
   UseMethod("filter_by_overlaps")
@@ -74,4 +78,28 @@ filter_by_non_overlaps.GenomicRanges <- function(x,y, maxgap = -1L, minoverlap =
                    ignore.strand = TRUE)
 }
 
-# TODO -- add in more variants here?
+#' @export
+#' @rdname ranges-filter-overlaps
+filter_by_overlaps_directed <- function(x,y, maxgap = -1L, minoverlap = 0L) {
+  UseMethod("filter_by_overlaps_directed")
+}
+
+#' @export
+filter_by_overlaps_directed.GenomicRanges <- function(x,y, maxgap = -1L, minoverlap = 0L) {
+  subsetByOverlaps(x,y, maxgap, minoverlap, ignore.strand = FALSE)
+}
+
+#' @export
+#' @rdname ranges-filter-overlaps
+filter_by_non_overlaps_directed <- function(x,y, maxgap, minoverlap) {
+  UseMethod("filter_by_non_overlaps_directed")
+}
+
+#' @export
+filter_by_non_overlaps_directed.GenomicRanges <- function(x,y, maxgap = -1L, minoverlap = 0L) {
+  subsetByOverlaps(x,y, maxgap, minoverlap,
+                   invert = TRUE,
+                   ignore.strand = FALSE)
+}
+
+# TODO -- add in more variants here?

tests/testthat/test-filter-by-overlap.R

@@ -0,0 +1,36 @@
+# test-filter-by-overlap.R
+
+test_that("filters by overlap & non-overlap work", {
+
+  subject_gr <- as_granges(data.frame(seqnames = rep("chr1", 3),
+                               start = c(1,1, 100),
+                               width = rep(100, 3),
+                               strand = c("+", "-", "+")))
+
+  query_gr <- as_granges(data.frame(seqnames = c("chr1", "chr2", "chr1"),
+                               start = c(20, 20, 150),
+                               width = c(10, 10, 10),
+                               strand = c("+", "-", "+")))
+
+  filter_directed_expected_gr <- as_granges(data.frame(seqnames = rep("chr1", 2),
+                               start = c(1, 100),
+                               width = rep(100, 2),
+                               strand = c("+", "+")))
+
+  filter_non_overlap_expected_gr <- GRanges(seqinfo = Seqinfo(seqnames = "chr1"))
+
+  filter_non_overlap_directed_expected_gr <- as_granges(
+    data.frame(seqnames = "chr1",
+                               start = c(1),
+                               width = rep(100),
+                               strand = c("-")
+               )
+    )
+
+  expect_identical(subject_gr, filter_by_overlaps(subject_gr, query_gr))
+  expect_identical(filter_directed_expected_gr, filter_by_overlaps_directed(subject_gr, query_gr))
+
+  expect_identical(filter_non_overlap_expected_gr, filter_by_non_overlaps(subject_gr, query_gr))
+  expect_identical(filter_non_overlap_directed_expected_gr, filter_by_non_overlaps_directed(subject_gr, query_gr))
+
+})