RnBeads (version 1.4.0)

computeDiffTab.default.site: computeDiffTab.site

Description

computes a difference table containing multiple difference measures, In the simple version the difference in means, quotients in means and a p-value for the comparison of two groups in a table are computed. This is computed for each row of the input table. The extended version contains additional columns

Usage

computeDiffTab.default.site(X, inds.g1, inds.g2, diff.method = rnb.getOption("differential.site.test.method"), paired = FALSE, adjustment.table = NULL, eps = 0.01)
computeDiffTab.extended.site(X, inds.g1, inds.g2, diff.method = rnb.getOption("differential.site.test.method"), paired = FALSE, adjustment.table = NULL, eps = 0.01, covg = NULL, covg.thres = rnb.getOption("filtering.coverage.threshold"))

Arguments

X

Matrix on which the difference measures are calculated for every row

inds.g1

column indices of group 1 members

inds.g2

column indices of group 2 members

diff.method

Method to determine p-values for differential methylation. Currently supported are "ttest" for a two-sided Welch t-test, "refFreeEWAS" for adjusting for cell mixtures, and "limma" for p-values resulting from linear modeling of the transformed beta values (M-values) and using techniques from expression microarray analysis employed in the limma package.

paired

should a paired a analysis be performed. If TRUE then inds.g1 and inds.g2 should have exactly the same length and should be order, such that the first element of inds.g1 corresponds to the first element of inds.g2 and so on.

adjustment.table

a table of variables to be adjusted for in the differential methylation test. Currently this is only supported for diff.method=="limma"

eps

Epsilon for computing quotients (avoid division by 0 by adding this value to denominator and enumerator before calculating the quotient)

covg

coverage information (should be NULL for disabled or of equal dimensions as X)

covg.thres

a coverage threshold

Value

a dataframe containing the following variables:

mean.g1

Mean of group 1

mean.g2

Mean of group 2

mean.diff

Difference in means

mean.quot.log2

log2 of the quotient of means

diffmeth.p.val

P-value (as determined by diff.method)

max.g1/max.g2

[extended version only] Group maxima

min.g1/min.g2

[extended version only] Group minima

sd.g1/sd.g2

[extended version only] Group standard deviations

min.diff

[extended version only] Minimum of 0 and single linkage difference between the groups

diffmeth.p.adj.fdr

[extended version only] FDR adjusted p-values

num.na.g1/num.na.g2

[extended version only] number of NA methylation values for groups 1 and 2 respectively

mean.covg.g1/mean.covg.g2

[extended version with coverage information only] mean coverage of groups 1 and 2 respectively

min.covg.g1/min.covg.g2

[extended version with coverage information only] minimum coverage of groups 1 and 2 respectively

max.covg.g1/max.covg.g2

[extended version with coverage information only] maximum coverage of groups 1 and 2 respectively

covg.thresh.nsamples.g1/2

[extended version with coverage information only] number of samples in group 1 and 2 respectively exceeding the coverage threshold for this site.

Examples

library(RnBeads.hg19)
data(small.example.object)
logger.start(fname=NA)
meth.mat <- meth(rnb.set.example)
sample.groups <- rnb.sample.groups(rnb.set.example)[[1]]
dm <- computeDiffTab.extended.site(meth.mat,sample.groups[[1]],sample.groups[[2]])
summary(dm)