dmrFinder: Find differentially methylated regions (DMRs)

Description

Find differentially methylated regions (DMRs) from tiling microarray data. If you want to adjust for covariates (including batch effects estimated using SVA) or if your covariate of interest is continuous, use the dmrFind function.

Usage

dmrFinder(eset=NULL, groups, p=NULL, l=NULL, chr=NULL, pos=NULL, pns=NULL,
	sdBins=NULL, controlIndex=NULL, controlProbes=NULL, Indexes=NULL, 
	filter=NULL, package=NULL, ws=7, verbose=TRUE, compare="all",  
	withinSampleNorm="loess", betweenSampleNorm="quantile", 
	cutoff=0.995, sortBy="ttarea", removeIf=expression(nprobes<3), paired="FALSE," pairs="NULL," dd="NULL," comps="NULL," comps.names="NULL)

Arguments

eset

a TilingFeatureSet

groups

a vector of group labels for the samples in eset

a matrix of percentage methylation values (scale: 0, 1). One column per sample

a matrix of methylation values (scale: -Inf, Inf), typically log-ratios.

chr

vector of chromosome labels for the probes in eset, p or l

pos

vector of chromosomal coordinates for the probes in eset, p or l

pns

vector of region names for the probes in eset, p or l

sdBins

not currently implemented

controlIndex

vector of indices of non-CpG control probes (which can be obtained using getControlIndex), to be passed on to the methp function if neither the p nor the l arguments are provided, in which case either this or the controlProbes argument must be provided.

controlProbes

names of probe containers corresponding to control probes (e.g., see values returned by getContainer), to be passed on to the methp function if neither the p nor the l arguments are provided, in which case either this or the controlIndex argument must be provided.

Indexes

not currently used

filter

smoothing window weights. See details

package

annotation package name

smoothing window size parameter. See details.

verbose

Verbose progress reporting

compare

the groups between which to find DMRs.

withinSampleNorm

within-sample normalization method. "loess" or "none"

betweenSampleNorm

between-sample normalization method. "quantile", "sqn" or "none"

cutoff

t-statistic cutoff used to identify probes as being in a DMR

sortBy

sort column for the DMR table. "area", "ttarea", "avg.diff", or "max.diff".

removeIf

expression indicating which DMRs to drop from the DMR tables that get returned. The negation of this is used as the subset argument to the subset function when it is called on the final DMR table before it is returned. If NULL, no DMRs will be subsetted out from the final table before it is returned. DMR table column names to use are listed below. E.g., to drop all DMRs with less than 4 probes, set removeIf=expression(nprobes

paired

if TRUE, do comparisons within pairs of samples. FALSE by default.

pairs

if paired=TRUE, this must be provided. a vector of pair identifiers for the samples in eset. values must be the same within pairs and different between pairs.

Ignore this argument.

COMPS

Ignore this argument.

COMPS.names

Ignore this argument.

Value

tabs

A list of DMR tables, one per comparison with columns:

chr: chromosome of DMR (bp)
start: start of DMR (bp)
end: end of DMR (bp)
p1: if paired=FALSE, and p!=NULL or l=NULL, average percentage methylation of all probes between start and end for group 1
p2: if paired=FALSE, and p!=NULL or l=NULL, average percentage methylation of all probes between start and end for group 2
m1: if paired=FALSE, p=NULL and l!=NULL, average methylation l (logit(percentage methylation) if l=NULL) of all probes between start and end for group 1
m2: if paired=FALSE, p=NULL and l!=NULL, average methylation l (logit(percentage methylation) if l=NULL) of all probes between start and end for group 2
regionName: name of the tiling region in which the DMR is found (These names come from the NDF file)
indexStart: index of first probe in DMR. This indexes the output of dmrFinder, *not* the input.
indexEnd: index of last probe in DMR. This indexes the output of dmrFinder, *not* the input.
nprobes: number of probes for the DMR, i.e., indexEnd-indexStart+1
diff: average percentage methylation difference within the DMR (i.e., column p1 - column p2), if p or eset arguments are provided. Otherwise, if only l argument is provided, it is the average difference in l (i.e, column m1 - column m2). Prior to version 2.0.1, if paired=TRUE, this was the average l (logit(percentage) methylation if l=NULL) difference within the DMR regardless of whether p or eset were provided.
maxdiff: maximum percentage methylation difference within the DMR, if p or eset arguments are provided. Otherwise, if only l argument is provided, it is the maximum difference in l. Prior to version 2.0.1, if paired=TRUE, this was the maximum l (logit(percentage) methylation if l=NULL) difference within the DMR regardless of whether p or eset were provided. Also prior to package version 2.0.1, this column was reported only in absolute value, however, post-version 2.0.1 the sign is retained.
area: nprobes x average difference
ttarea: nprobes x (average probe level t-statistic for between group difference)

p

A matrix of percentage methylation estimates (NOTE: the probe order may differ from that of the input p matrix since probes are sorted into chromosomal order)

l

This contains methylation log-ratios if they were passed to the function. Otherwise it contains logit-transformed percentage methylation estimates. (NOTE: the probe order may differ from that of the input l matrix since probes are sorted into chromosomal order)

chr

a vector of chromosomes corresponding to the rows of p and l

pos

a vector of positions corresponding to the rows of p and l

pns

a vector of probe region names corresponding to the rows of p and l

index

a vector identifying which subset of the input probes (i.e. which elements of the input chr, pos, and pns, and rows of the input p and/or l) were used to search for DMRs. The output objects (chr, pos, pns, p, l, etc) are this subset of probes from the input. Therefore, e.g., while tabs$indexStart:tabs$indexEnd indexes the elements or rows of the output objects for each DMR candidate in tabs, index[tabs$indexStart:tabs$indexEnd] indexes the elements or rows of the input objects.

gm

if paired=FALSE, group medians of the l matrix

DD

if paired=TRUE, a list of within-pair differences for each comparison

sMD

if paired=TRUE, a matrix of smoothed mean within-pair differences for each comparison

groups

a vector of group labels

args

the DMR finder parameter vector

comps

the vector of pairwise group comparisons

package

the array annotation package name

Details

This function finds differentially methylated regions (DMRs). The sortBy parameter can be used to sort the DMRs by area (# probes x average difference), t-statistic area (# probes x average t-statistic), average difference, or maximum difference.

Examples

Run this code

# See dmrFdr

Run the code above in your browser using DataLab

Description

Usage

Arguments

Value

Details

See Also

Examples