segmentMethCP: Perform segmentation on a 'MethCP' object.

In boyinggong/methcp: Differential methylation anlsysis for bisulfite sequencing data

Description

Perform CBS algorithm that segments the genome into similar levels of sigficance.

Usage

segmentMethCP(
    methcp.object, bs.object,
    region.test = c(
        "fisher", "stouffer", "weighted-variance", "weighted-coverage"),
    min.width = 2, sig.level = 0.01,
    presegment_dist = 600, BPPARAM = bpparam(), ...)

Arguments

methcp.object	a MethCP object.
bs.object	a BSseq object from the bsseq package.
region.test	The meta-analysis method used to create region-based test statistics.
min.width	the minimum width for the segments, which is used as termination rule for the segmentation algorithm.
sig.level	the significance level of the segments, which is used as termination rule for the segmentation algorithm.
presegment_dist	the maximum distance between cytosines for the presegmentation.
BPPARAM	An optional BiocParallelParam instance determining the parallel back-end to be used during evaluation, or a list of BiocParallelParam instances, to be applied in sequence for nested calls to BiocParallel functions. Default bpparam().
...	argument to be passed to segment function in DNAcopy package

Details

The MethCP object methcp.object can be generated from functions calcLociStat, calcLociStatTimeCourse, or methcpFromStat.

If region.test = "fisher", Fisher's combined probability test is used.

If region.test = stouffer Stouffer's test is applied.

If region.test = "weighted-variance" we use the variance of the test to combine per-cytosine based statistcis into a region-based statistic.

If region.test = "weighted-coverage" we use the coverage of the test to combine per-cytosine based statistcis into a region-based statistic.

Value

a MethCP object that is not segmented.

Examples

library(bsseq)
# Simulate a small dataset with 2000 cyotsine and 6 samples,
# 3 in the treatment group and 3 in the control group. The
# methylation ratio are generated using Binomial distribution
# with probability 0.3.
nC <- 2000
sim_cov <- rnbinom(6*nC, 5, 0.5) + 5
sim_M <- vapply(
    sim_cov, function(x) rbinom(1, x, 0.3), FUN.VALUE = numeric(1))
sim_cov <- matrix(sim_cov, ncol = 6)
sim_M <- matrix(sim_M, ncol = 6)
# methylation ratios in the DMRs in the treatment group are
# generated using Binomial(0.7)
DMRs <- c(600:622, 1089:1103, 1698:1750)
sim_M[DMRs, 1:3] <- vapply(
    sim_cov[DMRs, 1:3], function(x) rbinom(1, x, 0.7),
    FUN.VALUE = numeric(1))
# sample names
sample_names <- c(paste0("treatment", 1:3), paste0("control", 1:3))
colnames(sim_cov) <- sample_names
colnames(sim_M) <- sample_names

# create a bs.object
bs_object <- BSseq(gr = GRanges(
    seqnames = "Chr01", IRanges(start = (1:nC)*10, width = 1)),
    Cov = sim_cov, M = sim_M,
    sampleNames = sample_names)
DMRs_pos <- DMRs*10
methcp_obj1 <- calcLociStat(
    bs_object,
    group1 = paste0("treatment", 1:3),
    group2 = paste0("control", 1:3),
    test = "methylKit")
methcp_obj1 <- segmentMethCP(
    methcp_obj1, bs_object,
    region.test = "fisher")

boyinggong/methcp documentation built on April 25, 2021, 9 a.m.