dmrFinder: Finds differentially methylated regions for whole genome...

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

Finds differentially methylated regions for whole genome bisulfite sequencing data.

Description

Finds differentially methylated regions for whole genome bisulfite sequencing data. Essentially identifies regions of the genome where all methylation loci have an associated t-statistic that is beyond a (low, high) cutoff.

Usage

dmrFinder(bstat, cutoff = NULL, qcutoff = c(0.025, 0.975),
  maxGap=300, stat = "tstat.corrected", verbose = TRUE)

Arguments

bstat	An object of class BSseqStat or BSseqTstat.
cutoff	The cutoff of the t-statistics. This should be a vector of length two giving the (low, high) cutoff. If NULL, see qcutoff.
qcutoff	In case cutoff is NULL, compute the cutoff using these quantiles of the t-statistic.
maxGap	If two methylation loci are separated by this distance, break a possible DMR. This guarantees that the return DMRs have CpGs that are this distance from each other.
stat	Which statistic should be used?
verbose	Should the function be verbose?

Details

The workhorse function is BSmooth.tstat which sets up a t-statistic for a comparison between two groups.

Note that post-processing of these DMRs are likely to be necessary, filtering for example for length (or number of loci).

Value

A data.frame with columns

start, end, width, chr	genomic locations and width.
n	The number of methylation loci.
invdensity	Average length per loci.
group1.mean	The mean methylation level across samples and loci in 'group1'.
group2.mean	The mean methylation level across samples and loci in 'group2'.
meanDiff	The mean difference in methylation level; the difference between group1.mean and group2.mean.
idxStart, idxEnd, cluster	Internal use.
areaStat	The 'area' of the t-statistic; equal to the sum of the t-statistics for the individual methylation loci.
direction	either ‘hyper’ or ‘hypo’.
areaStat.corrected	Only present if column = "tstat.corrected", contains the area of the corrected t-statistics.

Author(s)

Kasper Daniel Hansen [e-mail].

References

KD Hansen, B Langmead, and RA Irizarry. BSmooth: from whole genome bisulfite sequencing reads to differentially methylated regions. Genome Biology (2012) 13:R83. doi:10.1186/gb-2012-13-10-r83.

See Also

BSmooth.tstat for the function constructing the input object, and BSseqTstat for its class. In the example below, we use BS.cancer.ex.tstat as the actual input object. Also see the package vignette(s) for a detailed example.

Examples

if(require(bsseqData)) {
  dmrs0 <- dmrFinder(BS.cancer.ex.tstat, cutoff = c(-4.6, 4.6), verbose = TRUE)
  dmrs <- subset(dmrs0, abs(meanDiff) > 0.1 & n >= 3)
}

kasperdanielhansen/bsseq documentation built on April 14, 2024, 2:19 a.m.