getSeq-methods

0th

Percentile

getSeq methods for BSgenome and XStringSet objects

getSeq methods for extracting a set of sequences (or subsequences) from a BSgenome or XStringSet object.

Keywords
manip
Usage
"getSeq"(x, names, start=NA, end=NA, width=NA, strand="+", as.character=FALSE)
"getSeq"(x, names)
Arguments
x
A BSgenome or XStringSet object. See the available.genomes function for how to install a genome.
names
When x is a BSgenome, names must be a character vector containing the names of the sequences in x where to get the subsequences from, or a GRanges object, or a GRangesList object, or a named RangesList object, or a named Ranges object. The RangesList or Ranges object must be named according to the sequences in x where to get the subsequences from.

If names is missing, then seqnames(x) is used. See ?BSgenome-class for details on how to get the lists of single sequences and multiple sequences (respectively) contained in a BSgenome object.

When x is a XStringSet object, names must be a character vector or GRanges object.

start, end, width
Vector of integers (eventually with NAs) specifying the locations of the subsequences to extract. These are not needed (and it's an error to supply them) when names is a GRanges, GRangesList, RangesList, or Ranges object.
strand
A vector containing "+"s or/and "-"s. This is not needed (and it's an error to supply it) when names is a GRanges or GRangesList object.
as.character
TRUE or FALSE. Should the extracted sequences be returned in a standard character vector?
...
Additional arguments. (Currently ignored.)
Details

L, the number of sequences to extract, is determined as follow:

• If names is a GRanges or Ranges object then L = length(names).
• If names is a GRangesList or RangesList object then L = length(unlist(names)).
• Otherwise, L is the length of the longest of names, start, end and width and all these arguments are recycled to this length. NAs and negative values in these 3 arguments are solved according to the rules of the SEW (Start/End/Width) interface (see ?solveUserSEW for the details).

If names is neither a GRanges or GRangesList object, then the strand argument is also recycled to length L.

Here is how the names passed to the names argument are matched to the names of the sequences in BSgenome object x. For each name in names:

• (1): If x contains a single sequence with that name then this sequence is used for extraction;
• (2): Otherwise the names of all the elements in all the multiple sequences are searched. If the names argument is a character vector then name is treated as a regular expression and grep is used for this search, otherwise (i.e. when the names are supplied via a higher level object like GRanges or GRangesList) then name must match exactly the name of the sequence. If exactly 1 sequence is found, then it is used for extraction, otherwise (i.e. if no sequence or more than 1 sequence is found) then an error is raised.

Value

Normally a DNAStringSet object (or character vector if as.character=TRUE).With the 2 following exceptions:
1. A DNAStringSetList object (or CharacterList object if as.character=TRUE) of the same shape as names if names is a GRangesList object.
2. A DNAString object (or single character string if as.character=TRUE) if L = 1 and names is not a GRanges, GRangesList, RangesList, or Ranges object.

Note

Be aware that using as.character=TRUE can be very inefficient when extracting a "big" amount of DNA sequences (e.g. millions of short sequences or a small number of very long sequences).

Note that the masks in x, if any, are always ignored. In other words, masked regions in the genome are extracted in the same way as unmasked regions (this is achieved by dropping the masks before extraction). See ?MaskedDNAString-class for more information about masked DNA sequences.

getSeq, available.genomes, BSgenome-class, DNAString-class, DNAStringSet-class, MaskedDNAString-class, GRanges-class, GRangesList-class, RangesList-class, Ranges-class, grep

Aliases
• getSeq
• getSeq,BSgenome-method
• getSeq,XStringSet-method
Examples
## ---------------------------------------------------------------------
## A. SIMPLE EXAMPLES
## ---------------------------------------------------------------------

## Load the Caenorhabditis elegans genome (UCSC Release ce2):
library(BSgenome.Celegans.UCSC.ce2)

## Look at the index of sequences:
Celegans

## Get chromosome V as a DNAString object:
getSeq(Celegans, "chrV")
## which is in fact the same as doing:
Celegans$chrV ## Not run: # ## Never try this: # getSeq(Celegans, "chrV", as.character=TRUE) # ## or this (even worse): # getSeq(Celegans, as.character=TRUE) # ## End(Not run) ## Get the first 20 bases of each chromosome: getSeq(Celegans, end=20) ## Get the last 20 bases of each chromosome: getSeq(Celegans, start=-20) ## --------------------------------------------------------------------- ## B. EXTRACTING SMALL SEQUENCES FROM DIFFERENT CHROMOSOMES ## --------------------------------------------------------------------- myseqs <- data.frame( chr=c("chrI", "chrX", "chrM", "chrM", "chrX", "chrI", "chrM", "chrI"), start=c(NA, -40, 8510, 301, 30001, 9220500, -2804, -30), end=c(50, NA, 8522, 324, 30011, 9220555, -2801, -11), strand=c("+", "-", "+", "+", "-", "-", "+", "-") ) getSeq(Celegans, myseqs$chr,
start=myseqs$start, end=myseqs$end)
getSeq(Celegans, myseqs$chr, start=myseqs$start, end=myseqs$end, strand=myseqs$strand)

## ---------------------------------------------------------------------
## C. USING A GRanges OBJECT
## ---------------------------------------------------------------------

gr1 <- GRanges(seqnames=c("chrI", "chrI", "chrM"),
ranges=IRanges(start=101:103, width=9))
gr1  # all strand values are "*"
getSeq(Celegans, gr1)  # treats strand values as if they were "+"

strand(gr1)[] <- "-"
getSeq(Celegans, gr1)

strand(gr1)[1] <- "+"
getSeq(Celegans, gr1)

strand(gr1)[2] <- "*"
if (interactive())
getSeq(Celegans, gr1)  # Error: cannot mix "*" with other strand values

gr2 <- GRanges(seqnames=c("chrM", "NM_058280_up_1000"),
ranges=IRanges(start=103:102, width=9))
gr2
if (interactive()) {
## Because the sequence names are supplied via a GRanges object, they
## are not treated as regular expressions:
getSeq(Celegans, gr2)  # Error: sequence NM_058280_up_1000 not found
}

## ---------------------------------------------------------------------
## D. USING A GRangesList OBJECT
## ---------------------------------------------------------------------

gr1 <- GRanges(seqnames=c("chrI", "chrII", "chrM", "chrII"),
ranges=IRanges(start=101:104, width=12),
strand="+")
gr2 <- shift(gr1, 5)
gr3 <- gr2
strand(gr3) <- "-"

grl <- GRangesList(gr1, gr2, gr3)
getSeq(Celegans, grl)

## ---------------------------------------------------------------------
## E. EXTRACTING A HIGH NUMBER OF RANDOM 40-MERS FROM A GENOME
## ---------------------------------------------------------------------

{
start <- lapply(seqnames(x),
function(name)
{
seqlength <- seqlengths(x)[name]
sample(seqlength - readlength + 1L,
seqlength * density,
replace=TRUE)
})
names <- rep.int(seqnames(x), elementNROWS(start))
ranges <- IRanges(start=unlist(start), width=readlength)
strand <- strand(sample(c("+", "-"), length(names), replace=TRUE))
gr <- GRanges(seqnames=names, ranges=ranges, strand=strand)
getSeq(x, gr)
}

## With a density of 1 read every 100 genome bases, the total number of
## extracted 40-mers is about 1 million:

## Notes:
## - The short sequences in 'rndreads' can be seen as the result of a
##   simulated high-throughput sequencing experiment. A non-realistic
##   one though because:
##     (a) It assumes that the underlying technology is perfect (the
##         generated reads have no technology induced errors).
##     (b) It assumes that the sequenced genome is exactly the same as
##         the reference genome.
##     (c) The simulated reads can contain IUPAC ambiguity letters only
##         because the reference genome contains them. In a real
##         high-throughput sequencing experiment, the sequenced genome
##         of course doesn't contain those letters, but the sequencer
##         can introduce them in the generated reads to indicate
##         ambiguous base-calling.
## - Those reads are coming from the plus and minus strands of the
##   chromosomes.
## - With a density of 0.01 and the reads being only 40-base long, the
##   average coverage of the genome is only 0.4 which is low. The total
##   number of reads is about 1 million and it takes less than 10 sec.
##   to generate them.
## - A higher coverage can be achieved by using a higher density and/or
##   longer reads. For example, with a density of 0.1 and 100-base reads
##   the average coverage is 10. The total number of reads is about 10
##   millions and it takes less than 1 minute to generate them.
## - Those reads could easily be mapped back to the reference by using
##   an efficient matching tool like matchPDict() for performing exact
##   matching (see ?matchPDict for more information). Typically, a
##   small percentage of the reads (4 to 5% in our case) will hit the
##   reference at multiple locations. This is especially true for such
##   short reads, and, in a lower proportion, is still true for longer
##   reads, even for reads as long as 300 bases.

## ---------------------------------------------------------------------
## F. SEE THE BSgenome CACHE IN ACTION
## ---------------------------------------------------------------------

options(verbose=TRUE)
first20 <- getSeq(Celegans, end=20)
first20
gc()
stopifnot(length(ls(Celegans@.seqs_cache)) == 0L)
## One more gc() call is needed in order to see the amount of memory in
## used after all the chromosomes have been removed from the cache:
gc()

Documentation reproduced from package BSgenome, version 1.40.1, License: Artistic-2.0

Community examples

Looks like there are no examples yet.