segmentSeq (version 2.6.0)

processAD: Processes an `alignmentData' or `alignmentMeth' object into a `segData' or `segMeth' object for segmentation.

Description

In order to discover segments of the genome with a high density of sequenced data, a `segData' object must be produced. This is an object containing a set of potential segments, together with the counts for each sample in each potential segment.

Usage

processAD(aD, gap = 200, squeeze = 0, filterProp = 0.1, strandSplit = FALSE,
verbose = TRUE, getCounts = FALSE, cl)

Arguments

aD

An alignmentData or alignmentMeth object.

gap

The maximum gap between aligned tags that should be allowed in constructing potential segments. Defaults to 200. See Details.

squeeze

If greater than zero, the minimum gap between aligned tags that should be allowed in constructing potential segments. See Details.

filterProp

If 'aD' is a alignmentMeth object and this is given, the minimum proportion of methylation at a base below which the base will be filtered out before constructing potential segments (but not during counting).

strandSplit

If TRUE, the data will be split by strand and segments will be constructed separately for each strand. Defaults to FALSE.

verbose

Should processing information be displayed? Defaults to TRUE.

getCounts

If TRUE, counts will be estimated for the constructed `segData' object. If FALSE, they will not, and must be estimated on the fly for further operations on the `segData' object, which is computationally wasteful but will substantially reduce the memory requirements.

cl

A SNOW cluster object, or NULL. See Details.

Value

A segData object.

Details

This function takes an alignmentData or alignmentMeth object and constructs a segData or segMeth object from it. The function creates a set of potential segments by looking for all locations on the genome where the start of a region of overlapping alignments (or, if `squeeze' is non-zero, those alignments separated by no more than `squeeze'.) exists in the alignmentData object. A potential segment then exists from this start point to the end of all regions of overlapping alignments such that there is no region in the segment of at least length `gap' where no tag aligns. The number of potential segments can therefore be increased by increasing this limit, or (usually more usefully) decreased by decreasing this limit in order to save computational effort.

A 'cluster' object (package: snow) is recommended for parallelisation of this function when using large data sets. Passing NULL to this variable will cause the function to run in non-parallel mode.

See Also

getCounts, which produces the count data for each potential segment. heuristicSeg and classifySeg, which segment the genome based on the segData object produced by this function segData alignmentData

Examples

# Define the chromosome lengths for the genome of interest.

chrlens <- c(2e6, 1e6)

# Define the files containing sample information.

datadir <- system.file("extdata", package = "segmentSeq")
libfiles <- c("SL9.txt", "SL10.txt", "SL26.txt", "SL32.txt")

# Establish the library names and replicate structure.

libnames <- c("SL9", "SL10", "SL26", "SL32")
replicates <- c(1,1,2,2)

# Process the files to produce an `alignmentData' object.

alignData <- readGeneric(file = libfiles, dir = datadir, replicates =
replicates, libnames = libnames, chrs = c(">Chr1", ">Chr2"), chrlens =
chrlens, gap = 100)

# Process the alignmentData object to produce a `segData' object.

sD <- processAD(alignData, gap = 100, cl = NULL)