Put (virtual) tiles on a given genome
tileGenome returns a set of genomic regions that form a
partitioning of the specified genome. Each region is called a "tile".
tileGenome(seqlengths, ntile, tilewidth, cut.last.tile.in.chrom=FALSE)
seqlengths |
Either a named numeric vector of chromosome lengths or a Seqinfo
object. More precisely, if a named numeric vector, it must have a length
>= 1, cannot contain NAs or negative values, and cannot have duplicated
names. If a Seqinfo object, then it's first replaced with the
vector of sequence lengths stored in the object (extracted from the object
with the |
ntile |
The number of tiles to generate. |
tilewidth |
The desired tile width. The effective tile width might be slightly different but is guaranteed to never be more than the desired width. |
cut.last.tile.in.chrom |
Whether or not to cut the last tile in each chromosome.
This is set to |
If cut.last.tile.in.chrom is FALSE (the default),
a GRangesList object with one list element per tile, each of
them containing a number of genomic ranges equal to the number of
chromosomes it overlaps with. Note that when the tiles are small (i.e.
much smaller than the chromosomes), most of them only overlap with a
single chromosome.
If cut.last.tile.in.chrom is TRUE, a GRanges
object with one element (i.e. one genomic range) per tile.
H. Pagès, based on a proposal by M. Morgan
genomicvars for an example of how to compute the binned average of a numerical variable defined along a genome.
GRangesList and GRanges objects.
Seqinfo objects and the seqlengths getter.
IntegerList objects.
Views objects.
## ---------------------------------------------------------------------
## A. WITH A TOY GENOME
## ---------------------------------------------------------------------
seqlengths <- c(chr1=60, chr2=20, chr3=25)
## Create 5 tiles:
tiles <- tileGenome(seqlengths, ntile=5)
tiles
elementNROWS(tiles) # tiles 3 and 4 contain 2 ranges
width(tiles)
## Use sum() on this IntegerList object to get the effective tile
## widths:
sum(width(tiles)) # each tile covers exactly 21 genomic positions
## Create 9 tiles:
tiles <- tileGenome(seqlengths, ntile=9)
elementNROWS(tiles) # tiles 6 and 7 contain 2 ranges
table(sum(width(tiles))) # some tiles cover 12 genomic positions,
# others 11
## Specify the tile width:
tiles <- tileGenome(seqlengths, tilewidth=20)
length(tiles) # 6 tiles
table(sum(width(tiles))) # effective tile width is <= specified
## Specify the tile width and cut the last tile in each chromosome:
tiles <- tileGenome(seqlengths, tilewidth=24,
cut.last.tile.in.chrom=TRUE)
tiles
width(tiles) # each tile covers exactly 24 genomic positions, except
# the last tile in each chromosome
## Partition a genome by chromosome ("natural partitioning"):
tiles <- tileGenome(seqlengths, tilewidth=max(seqlengths),
cut.last.tile.in.chrom=TRUE)
tiles # one tile per chromosome
## sanity check
stopifnot(all.equal(setNames(end(tiles), seqnames(tiles)), seqlengths))
## ---------------------------------------------------------------------
## B. WITH A REAL GENOME
## ---------------------------------------------------------------------
library(BSgenome.Scerevisiae.UCSC.sacCer2)
tiles <- tileGenome(seqinfo(Scerevisiae), ntile=20)
tiles
tiles <- tileGenome(seqinfo(Scerevisiae), tilewidth=50000,
cut.last.tile.in.chrom=TRUE)
tiles
## ---------------------------------------------------------------------
## C. AN APPLICATION: COMPUTE THE BINNED AVERAGE OF A NUMERICAL VARIABLE
## DEFINED ALONG A GENOME
## ---------------------------------------------------------------------
## See '?genomicvars' for an example of how to compute the binned
## average of a numerical variable defined along a genome.Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.