BSseq: The constructor function for BSseq objects.

In kasperdanielhansen/bsseq: Analyze, manage and store whole-genome methylation data

The constructor function for BSseq objects.

Description

The constructor function for BSseq objects.

Usage

BSseq(M = NULL, Cov = NULL, Filtered = NULL, coef = NULL, se.coef = NULL,
  trans = NULL, parameters = NULL, pData = NULL, gr = NULL,
  pos = NULL, chr = NULL, sampleNames = NULL, rmZeroCov = FALSE)

Arguments

M	A matrix-like object of methylation evidence (see 'Details' below).
Cov	A matrix-like object of coverage (see 'Details' below)).
Filtered	A matrix-like object of ambiguous modification bases obtained from modbam2bed.
coef	A matrix-like object of smoothing estimates (see 'Details' below).
se.coef	A matrix-like object of smoothing standard errors (see 'Details' below).
trans	A smoothing transformation.
parameters	A list of smoothing parameters.
pData	An data.frame or DataFrame.
sampleNames	A vector of sample names.
gr	An object of type GRanges.
pos	A vector of locations.
chr	A vector of chromosomes.
rmZeroCov	Should genomic locations with zero coverage in all samples be removed.

Details

The 'M', 'Cov', 'coef', and 'se.coef' matrix-like objects will be coerced to DelayedMatrix objects; see ?DelayedArray::DelayedMatrix for the full list of supported matrix-like objects. We recommend using matrix objects for in-memory storage of data and HDF5Matrix for on-disk storage of data.

Genomic locations are specified either through gr or through chr and pos but not both. There should be the same number of genomic locations as there are rows in the M and Cov matrix.

The argument rmZeroCov may be useful in order to reduce the size of the return object be removing methylation loci with zero coverage.

In case one or more methylation loci appears multiple times, the M and Cov matrices are summed over rows linked to the same methylation loci. See the example below.

Users should never have to specify coef, se.coef, trans, and parameters, this is for internal use (they are added by BSmooth).

phenoData is a way to specify pheno data (as known from the ExpressionSet and eSet classes), at a minimum sampleNames should be given (if they are not present, the function uses col.names(M)).

Value

An object of class BSseq.

Author(s)

Kasper Daniel Hansen [e-mail]

See Also

BSseq

Examples

  M <- matrix(0:8, 3, 3)
  Cov <- matrix(1:9, 3, 3)
  BS1 <- BSseq(chr = c("chr1", "chr2", "chr1"), pos = c(1,2,3),
               M = M, Cov = Cov, sampleNames = c("A","B", "C"))
  BS1
  BS2 <- BSseq(chr = c("chr1", "chr1", "chr1"), pos = c(1,1,1),
               M = M, Cov = Cov, sampleNames = c("A","B", "C"))
  BS2

  #-------------------------------------------------------------------------------
  # An example using a HDF5Array-backed BSseq object
  #

  library(HDF5Array)
  hdf5_M <- realize(M, "HDF5Array")
  hdf5_Cov <- realize(Cov, "HDF5Array")
  hdf5_BS1 <- BSseq(chr = c("chr1", "chr2", "chr1"),
                    pos = c(1, 2, 3),
                    M = hdf5_M,
                    Cov = hdf5_Cov,
                    sampleNames = c("A", "B", "C"))
  hdf5_BS1
  hdf5_BS2 <- BSseq(chr = c("chr1", "chr1", "chr1"),
                    pos = c(1, 1, 1),
                    M = hdf5_M,
                    Cov = hdf5_Cov,
                    sampleNames = c("A", "B", "C"))
  hdf5_BS2

kasperdanielhansen/bsseq documentation built on Dec. 17, 2024, 12:47 p.m.