enrich.mrf: Detection of enriched and differentially bound regions for fitting results of `mrf` and `mrf.joint`.

Description

enrich.mrf returns the enriched regions or differentially bound regions using the mrf or the mrf.joint model, by controlling a given FDR level. enrich.mrf also calculates the IP efficiencies for each experiment.

Usage

enrich.mrf(object, analysis = "joint", differential = FALSE, 
    diff.vec = NULL, cr = 0.05, crdiff = 0.05)

Arguments

object

The output of mrf if analysis="separate" or of mrf.joint if analysis="joint".

analysis

A character variable. Default value is "joint" and the object should be the output of mrf.joint. If analysis="separate", then the object should be the output of mrf.

differential

A logical variable. If TRUE, the function will compute the posterior probability of differential binding of any two experiments or two conditions, as specified by diff.vec. Default value is FALSE.

diff.vec

A numeric vector. If differential = TRUE, diff.vec must be given to show which experiments are to be used in the comparison. At the moment, this is restricted to two conditions (e.g. two proteins at the same time point), so the value for diff.vec should b

A numeric variable. The level of FDR for identifying the enriched regions.

crdiff

A numeric variable. The level of FDR for identifying the differentially bound regions.

Value

enrichThe list of enriched regions for each condition at the chosen FDR. Note that there is only one list of enriched regions for replicates, if a joint model is used.
diffenrich1The list of regions bound only by condition 1.
diffenrich2The list of regions bound only by condition 2.
XA n x p matrix of enrichment for each region and each condition, at the given FDR cutoff (1: enriched, 0: not-enriched).
diffprob1A n-dimensional vector of posterior probabilities of differential binding for the two conditions under study; diffprob0=1-diffprob1.
diffX1A n-dimensional index of regions bound only by condition 1 (0: not bound, 1: bound).
diffX2A n-dimensional index of regions bound only by condition 2.
IPEA p-dimensional vector of estimated IP efficiency values for each experiment.

References

Bao et al. Joint modelling of ChIP-seq data via a Markov random field model, Biostatistics 2014, 15(2):296-310 DOI:10.1093/biostatistics/kxt047.

Examples

Run this code

data(p300cbp.200bp)
exp.label=c("CBPT0", "CBPT301", "CBPT302", "p300T0", 
    "p300T301", "p300T302", "WangCBP", "Wangp300")
CBPT0=list()
CBPT0$region=p300cbp.200bp$region[200001:210000,]
CBPT0$count=p300cbp.200bp$count[200001:210000,1]

NBfit.simple<-mrf(CBPT0, method="NB", exp.label=exp.label[1])
enrich.mrf.simple<-enrich.mrf(NBfit.simple, analysis="separate")

## Joint analysis combining technical replicates 
##   (CBPT301,CBPT302) and (p300T301, p300T302)
p300cbp.mrf<-mrf.joint(p300cbp.200bp, method="NB", 
    rep.vec=c(1, 2,2,3,4,4, 5,6), p.vec=c(1,2,2, 3,4,4, 5,6), exp.label=exp.label)

## Enriched regions for all conditions and differentially bound 
##   regions for CBP and p300 at T30 using a 5p300cbp.enrich<-enrich.mrf(p300cbp.mrf, differential=TRUE, diff.vec=c(0,1,1,0,2,2,0,0))