Package 'ChIPComp'

Detect differential binding sites for ChIP sequencing data

Description

ChIPComp is an R library performing the differential binding analysis for ChIP-seq count data. Compared with other similar packages (DBChIP, DIME), ChIPComp considers the control samples in the process of detecting the differential binding sites. Extensive simulation results showed that ChIPComp performs favorably compared to DBChIP and DIME when the control samples are ignored. ChIPComp only works for two group comparison at this time, that is, to detect the differential binding sites for one transcription factor(histone) between two conditions (cell lines). We plan to extend the functionalities and make it work for more general experimental designs in the near future.

Author(s)

Hao Wu <[email protected]>, Li Chen <[email protected]>

---

Perform hypothesis testing to detect differential binding sites

Description

Perform hypothesis testing to detect differential binding sites

Usage

ChIPComp(countSet,A,threshold=1)

Arguments

countSet	A ChIPComp object.
A	User-specified regions to fit the model. It is a bed file with three columns, named ("chr","start","end"), could be separated by space or tab.
threshold	User specified posterior probability threshold. Default is 1.

Value

A object ChIPComp contains Column chr,start,end are the binding site genomic coordinate; Column ip_c(\#condition)_r(\#replicate) indicates ChIP counts in \#replicate in \#condition; Column ct_c(\#condition)_r(\#replicate) indicates smoothing control counts in \#replicate in \#condition; Column commonPeak 1s indicate common binding sites; Column prob.post is the posterior probability for each binding site. Column pvalue.wald is the pvalue of wald test for each binding site.

Author(s)

Hao Wu<[email protected]>, Li Chen <[email protected]>

Examples

data(seqData)
	seqData=ChIPComp(seqData)

---

make configurations for experimental design written in csv sheet

Description

Make a list with two elements. The first element is a data frame containing two group comparison study information. The second element is the design matrix.

Usage

makeConf(sampleSheet)

Arguments

sampleSheet

A csv sheet represents ChIP experiments design. It contains 6 columns,sampleID,condition,factor,ipReads,ctReads,peaks. condition refers to treatment condition or cell line; factor refers to transcription factor or histone modification; ipReads is the ChIP sequence data in bam or bed format; ctReads is the control sequence data in bam or bed format; peaks is the called peaks from existing peak-calling software.

Value

A list with two elements. The first element is a data frame containing two group comparison study information. The second element is the design matrix.

Author(s)

Hao Wu<[email protected]>, Li Chen <[email protected]>

Examples

confs=makeConf(system.file("extdata", "conf.csv", package="ChIPComp"))
	conf=confs$conf
	design=confs$design

---

make differential binding sites data frame

Description

This is an utility function to create a data frame. The data frame contains binding sites merged by peaks from two conditions, count ChIP read counts, smoothing control counts for each candidate region, and indicate the common peaks from two conditions.

Usage

makeCountSet(conf,design,filetype,species,peak.center=FALSE,peak.ext=0,binsize=50,mva.span=c(1000,5000,10000))

Arguments

conf	A data frame that represents the ChIP experiments information. It contains 6 columns,sampleID,condition,factor,ipReads,ctReads,peaks. condition refers to treatment condition or cell line; factor refers to transcription factor or histone modification; ipReads is the ChIP sequence data in bam or bed format; ctReads is the control sequence data in bam or bed format; peaks is the called peaks from existing peak-calling software.
design	Two column design matrix. The number of rows equals number of ChIP samples from two conditions. The first column are all 1s, which indicates intercept in regression model. The second column are 1s for one condition and 0s for another condition.
filetype	Two sequence file types are supported (bed or bam).
species	Two species are supported (hg19 or mm9). Other species are supported by specifying other.
peak.center	This argument is coupled with peak.ext. Default is FALSE. The argument is used when centered regions of peaks are more of interest.
peak.ext	This argument is coupled with peak.center. Default is 0.
binsize	binsize in bp to calculate the smooth local lambda in poisson distribution. The default is 50bp.
mva.span	1 kb, 5 kb or 10 kb window centered at the peak location in the control sample.

Value

A object ChIPComp. Column chr,start,end are the binding site genomic coordinate; Column ip_c(\#condition)_r(\#replicate) indicates the ChIP counts in \#replicate in \#condition; Column ct_c(\#condition)_r(\#replicate) indicates the smoothing control counts in \#replicate in \#condition; Column commonPeak indicates the common binding sites.

Examples

conf=data.frame(
		SampleID=1:4,
		condition=c("Helas3","Helas3","K562","K562"),
		factor=c("H3k27ac","H3k27ac","H3k27ac","H3k27ac"),
		ipReads=system.file("extdata",c("Helas3.ip1.bed","Helas3.ip2.bed","K562.ip1.bed","K562.ip2.bed"),package="ChIPComp"),
		ctReads=system.file("extdata",c("Helas3.ct.bed","Helas3.ct.bed","K562.ct.bed","K562.ct.bed"),package="ChIPComp"),
		peaks=system.file("extdata",c("Helas3.peak.bed","Helas3.peak.bed","K562.peak.bed","K562.peak.bed"),package="ChIPComp")
	)
	conf$condition=factor(conf$condition)
  conf$factor=factor(conf$factor)
	design=as.data.frame(lapply(conf[,c("condition","factor")],as.numeric))-1
	design=as.data.frame(model.matrix(~condition,design))
	countSet=makeCountSet(conf,design,filetype="bed", species="hg19",binsize=1000)

---

plot correlation between log ChIP read counts and smoothing control counts in common binding sites.

Description

plot correlation between log ChIP counts and smoothing control counts in common binding sites.

Usage

## S3 method for class 'ChIPComp'
plot(x,...)

Arguments

x	A ChIPComp object.
...	Other graphical parameters to plot

Value

Plot the correlation between ChIP sample and control sample

Author(s)

Hao Wu<[email protected]>, Li Chen <[email protected]>

Examples

data(seqData)
	plot(seqData)

---

Print top ranked differential binding sites

Description

Print top differential binding sites ranked by posterior probability in a decreasing order.

Usage

## S3 method for class 'ChIPComp'
print(x,topK=10,...)

Arguments

x	A ChIPComp object.
topK	top K differential binding sites. Default is 10.
...	Other parameters to print

Value

Print differential binding sites ranked by posterior probability

Author(s)

Hao Wu<[email protected]>, Li Chen <[email protected]>

Examples

data(seqData)
	seqData=ChIPComp(seqData)
	print(seqData)

---

A ChIPComp object.

Description

The object is sampled from 50 common binding sites between Helas3 and K562 cell lines for H3K27ac and 5 unique binding sites for each cell line.

Usage

data(seqData)

Value

A "ChIPComp" class object

Examples

data(seqData)

---