Added a findOverlap method for character vectors.

R/findOverlaps-methods.R

@@ -223,6 +223,73 @@ setMethod("findOverlaps", c("GenomicRanges", "GRangesList"),
     }
 )
 
+setMethod("findOverlaps", c("GenomicRanges", "character"),
+    function(query, subject, maxgap=-1L, minoverlap=0L,
+             type=c("any", "start", "end", "within", "equal"),
+             select=c("all", "first", "last", "arbitrary"),
+             ignore.strand=FALSE)
+    {
+        ans <- findMatches(seqnames(query), subject)
+        selectHits(ans, select=match.arg(select))
+    }
+)
+
+setMethod("findOverlaps", c("character", "GenomicRanges"),
+    function(query, subject, maxgap=-1L, minoverlap=0L,
+             type=c("any", "start", "end", "within", "equal"),
+             select=c("all", "first", "last", "arbitrary"),
+             ignore.strand=FALSE)
+    {
+        ans <- findMatches(query, seqnames(subject))
+        selectHits(ans, select=match.arg(select)) 
+    }
+)
+
+setMethod("findOverlaps", c("GenomicRangesList", "character"),
+    function(query, subject, maxgap=-1L, minoverlap=0L,
+             type=c("any", "start", "end", "within", "equal"),
+             select=c("all", "first", "last", "arbitrary"),
+             ignore.strand=FALSE)
+    {
+        sn <- unique(seqnames(query))
+
+        type <- match.arg(type)
+        if (type == "within") {
+            remap <- which(lengths(sn) == 1L)
+            ans <- findMatches(unlist(sn[remap]), subject)
+        } else {
+            ans <- findMatches(unlist(sn), subject)
+            remap <- rep(seq_along(sn), lengths(sn))
+        } 
+
+        ans2 <- Hits(from = unname(remap[queryHits(ans)]), to = subjectHits(ans), nLnode = length(query), nRnode = length(subject))
+        ans2 <- unique(ans2)
+        selectHits(ans2, select=match.arg(select))
+    }
+)
+
+setMethod("findOverlaps", c("character", "GenomicRangesList"),
+    function(query, subject, maxgap=-1L, minoverlap=0L,
+             type=c("any", "start", "end", "within", "equal"),
+             select=c("all", "first", "last", "arbitrary"),
+             ignore.strand=FALSE)
+    {
+        sn <- unique(seqnames(subject))
+
+        type <- match.arg(type)
+        if (type == "within") {
+            remap <- which(lengths(sn) == 1L)
+            ans <- findMatches(query, as.character(unlist(sn[remap])))
+        } else {
+            ans <- findMatches(query, as.character(unlist(sn))) # TODO: fix bug in findMatches with seqnames as the second argument.
+            remap <- rep(seq_along(sn), lengths(sn))
+        } 
+
+        ans2 <- Hits(from = queryHits(ans), to = unname(remap[subjectHits(ans)]), nLnode = length(query), nRnode = length(subject))
+        ans2 <- unique(ans2)
+        selectHits(ans2, select=match.arg(select))
+    }
+)
 
 ### =========================================================================
 ### findOverlaps-based methods

inst/unitTests/test_findOverlaps-methods.R

@@ -332,3 +332,41 @@ test_poverlaps <- function() {
                      GRanges("chr1:16-20"))
     checkIdentical(ans, Rle(c(FALSE, TRUE)))
 }
+
+test_findOverlaps_character <- function() {
+    gr <- GRanges(c("chr1:11-15", "chr1:5-100", "chr2:10-11"))
+
+    ans <- findOverlaps(gr, "chr1")
+    checkIdentical(queryHits(ans), which(seqnames(gr)=="chr1"))
+
+    ans <- findOverlaps(gr, c("chr2", "chr1"))
+    checkIdentical(as.character(seqnames(gr)[queryHits(ans)]), c("chr2", "chr1")[subjectHits(ans)])
+
+    # Works in the other direction.
+    ans <- findOverlaps("chr2", gr)
+    checkIdentical(subjectHits(ans), which(seqnames(gr)=="chr2"))
+
+    # Works for GRLs.
+    gr <- GRanges(c("chr1:11-15", "chr1:5-100", "chr2:1-10", "chr2:10-11"))
+    grl <- split(gr, c(1,2,2,3))
+    ans <- findOverlaps(grl, "chr2")
+    checkIdentical(queryHits(ans), unname(which(any(seqnames(grl) %in% "chr2"))))
+
+    ans <- findOverlaps(grl, c("chr2", "chr1"))
+    checkIdentical(queryHits(ans)[subjectHits(ans)==1L], unname(which(any(seqnames(grl) %in% "chr2"))))
+    checkIdentical(queryHits(ans)[subjectHits(ans)==2L], unname(which(any(seqnames(grl) %in% "chr1"))))
+
+    ans <- findOverlaps(grl, c("chr2", "chr1"), type="within")
+    checkIdentical(queryHits(ans)[subjectHits(ans)==1L], unname(which(all(seqnames(grl) %in% "chr2"))))
+    checkIdentical(queryHits(ans)[subjectHits(ans)==2L], unname(which(all(seqnames(grl) %in% "chr1"))))
+
+    # Same result in the other direction.
+    ans2 <- findOverlaps(c("chr2", "chr1"), grl)
+    checkIdentical(subjectHits(ans2)[queryHits(ans2)==1L], unname(which(any(seqnames(grl) %in% "chr2"))))
+    checkIdentical(subjectHits(ans2)[queryHits(ans2)==2L], unname(which(any(seqnames(grl) %in% "chr1"))))
+
+    ans2 <- findOverlaps(c("chr2", "chr1"), grl, type="within")
+    checkIdentical(subjectHits(ans2)[queryHits(ans2)==1L], unname(which(all(seqnames(grl) %in% "chr2"))))
+    checkIdentical(subjectHits(ans2)[queryHits(ans2)==2L], unname(which(all(seqnames(grl) %in% "chr1"))))
+}
+

man/findOverlaps-methods.Rd

@@ -54,6 +54,7 @@
 \arguments{
   \item{query, subject}{
     A \link{GRanges} or \link{GRangesList} object.
+    Either argument may also be a character vector of sequance names.
   }
   \item{maxgap, minoverlap, type}{
     See \code{?\link[IRanges]{findOverlaps}} in the \pkg{IRanges} package
@@ -97,6 +98,16 @@
   For \code{type="equal"} with GRangesList objects, \code{query[[i]]}
   matches \code{subject[[j]]} iff for each range in \code{query[[i]]}
   there is an identical range in \code{subject[[j]]}, and vice versa.
+
+  When either the query or subject is a character vector, the values are
+  interpreted as the sequence names. A range is considered to \dQuote{overlap}
+  with a sequence name if it lies on that sequence. \code{maxgap} and
+  \code{minoverlap} are ignored. \code{type} is ignored for \link{GRanges}, and
+  only \code{type="any"} and \code{type="within"} are used for
+  \link{GRangesList} inputs (the latter requiring all ranges to lie on the
+  specified sequence to be considered as an overlap). This method allows users
+  to conveniently identify ranges on certain chromosomes without worrying about
+  whether the ranges are stored in a \link{GRanges} or \link{GRangesList}. 
 }
 
 \value{