ChIPseeker: an R package for ChIP peak Annotation, Comparison and Visualization

In ChIPseeker: ChIPseeker for ChIP peak Annotation, Comparison, and Visualization

knitr::opts_chunk$set(tidy         = FALSE,
                      warning      = FALSE,
                      message      = FALSE)

CRANpkg <- function (pkg) {
    cran <- "https://CRAN.R-project.org/package"
    fmt <- "[%s](%s=%s)"
    sprintf(fmt, pkg, cran, pkg)
}

Biocpkg <- function (pkg) {
    sprintf("[%s](http://bioconductor.org/packages/%s)", pkg, pkg)
}

Biocannopkg <- Biocpkg

library(GenomicFeatures)
library(GenomicRanges)
library(TxDb.Hsapiens.UCSC.hg19.knownGene)
library(org.Hs.eg.db)
library(ggplot2)
library(clusterProfiler)
library(ReactomePA)
library(ChIPseeker)

Abstract

ChIPseeker is an R package for annotating ChIP-seq data analysis. It supports annotating ChIP peaks and provides functions to visualize ChIP peaks coverage over chromosomes and profiles of peaks binding to TSS regions. Comparison of ChIP peak profiles and annotation are also supported. Moreover, it supports evaluating significant overlap among ChIP-seq datasets. Currently, ChIPseeker contains 17,000 bed file information from GEO database. These datasets can be downloaded and compare with user's own data to explore significant overlap datasets for inferring co-regulation or transcription factor complex for further investigation.

Citation

If you use r Biocpkg("ChIPseeker")[@yu_chipseeker_2015] in published research, please cite:

G Yu, LG Wang, QY He. ChIPseeker: an R/Bioconductor package for ChIP peak annotation, comparison and visualization. Bioinformatics 2015, 31(14):2382-2383. doi:10.1093/bioinformatics/btv145

Introduction

Chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-seq) has become standard technologies for genome wide identification of DNA-binding protein target sites. After read mappings and peak callings, the peak should be annotated to answer the biological questions. Annotation also create the possibility of integrating expression profile data to predict gene expression regulation. r Biocpkg("ChIPseeker")[@yu_chipseeker_2015] was developed for annotating nearest genes and genomic features to peaks.

ChIP peak data set comparison is also very important. We can use it as an index to estimate how well biological replications are. Even more important is applying to infer cooperative regulation. If two ChIP seq data, obtained by two different binding proteins, overlap significantly, these two proteins may form a complex or have interaction in regulation chromosome remodelling or gene expression. r Biocpkg("ChIPseeker")[@yu_chipseeker_2015] support statistical testing of significant overlap among ChIP seq data sets, and incorporate open access database GEO for users to compare their own dataset to those deposited in database. Protein interaction hypothesis can be generated by mining data deposited in database. Converting genome coordinations from one genome version to another is also supported, making this comparison available for different genome version and different species.

Several visualization functions are implemented to visualize the coverage of the ChIP seq data, peak annotation, average profile and heatmap of peaks binding to TSS region.

Functional enrichment analysis of the peaks can be performed by my Bioconductor packages r Biocpkg("DOSE")[@yu_dose_2015], r Biocpkg("ReactomePA")[@yu_reactomepa_2016], r Biocpkg("clusterProfiler")[@yu_clusterprofiler_2012].

## loading packages
library(ChIPseeker)
library(TxDb.Hsapiens.UCSC.hg19.knownGene)
txdb <- TxDb.Hsapiens.UCSC.hg19.knownGene
library(clusterProfiler)

ChIP profiling

The datasets CBX6 and CBX7 in this vignettes were downloaded from GEO (GSE40740)[@pemberton_genome-wide_2014] while ARmo_0M, ARmo_1nM and ARmo_100nM were downloaded from GEO (GSE48308)[@urbanucci_overexpression_2012] . r Biocpkg("ChIPseeker") provides readPeakFile to load the peak and store in GRanges object.

files <- getSampleFiles()
print(files)
peak <- readPeakFile(files[[4]])
peak

ChIP peaks coverage plot

After peak calling, we would like to know the peak locations over the whole genome, covplot function calculates the coverage of peak regions over chromosomes and generate a figure to visualize. GRangesList is also supported and can be used to compare coverage of multiple bed files.

covplot(peak, weightCol="V5")

covplot(peak, weightCol="V5", chrs=c("chr17", "chr18"), xlim=c(4.5e7, 5e7))

Profile of ChIP peaks binding to TSS regions

First of all, for calculating the profile of ChIP peaks binding to TSS regions, we should prepare the TSS regions, which are defined as the flanking sequence of the TSS sites. Then align the peaks that are mapping to these regions, and generate the tagMatrix.

## promoter <- getPromoters(TxDb=txdb, upstream=3000, downstream=3000)
## tagMatrix <- getTagMatrix(peak, windows=promoter)
##
## to speed up the compilation of this vignettes, we use a precalculated tagMatrix
data("tagMatrixList")
tagMatrix <- tagMatrixList[[4]]

In the above code, you should notice that tagMatrix is not restricted to TSS regions. The regions can be other types that defined by the user.

Heatmap of ChIP binding to TSS regions

tagHeatmap(tagMatrix, xlim=c(-3000, 3000), color="red")

r Biocpkg("ChIPseeker") provide a one step function to generate this figure from bed file. The following function will generate the same figure as above.

peakHeatmap(files[[4]], TxDb=txdb, upstream=3000, downstream=3000, color="red")

Average Profile of ChIP peaks binding to TSS region

plotAvgProf(tagMatrix, xlim=c(-3000, 3000),
            xlab="Genomic Region (5'->3')", ylab = "Read Count Frequency")

The function plotAvgProf2 provide a one step from bed file to average profile plot. The following command will generate the same figure as shown above.

plotAvgProf2(files[[4]], TxDb=txdb, upstream=3000, downstream=3000,
             xlab="Genomic Region (5'->3')", ylab = "Read Count Frequency")

Confidence interval estimated by bootstrap method is also supported for characterizing ChIP binding profiles.

plotAvgProf(tagMatrix, xlim=c(-3000, 3000), conf = 0.95, resample = 1000)

Profile of ChIP peaks binding to start site of Exon/Intron

Referring to the issue #16, we developed getBioRegion function to support centering all peaks to the start region of Exon/Intron. Users can also create heatmap or average profile of ChIP peaks binding to these regions.

Peak Annotation

peakAnno <- annotatePeak(files[[4]], tssRegion=c(-3000, 3000),
                         TxDb=txdb, annoDb="org.Hs.eg.db")

Peak Annotation is performed by annotatePeak. User can define TSS (transcription start site) region, by default TSS is defined from -3kb to +3kb. The output of annotatePeak is csAnno instance. r Biocpkg("ChIPseeker") provides as.GRanges to convert csAnno to GRanges instance, and as.data.frame to convert csAnno to data.frame which can be exported to file by write.table.

TxDb object contained transcript-related features of a particular genome. Bioconductor provides several package that containing TxDb object of model organisms with multiple commonly used genome version, for instance r Biocannopkg("TxDb.Hsapiens.UCSC.hg38.knownGene"), r Biocannopkg("TxDb.Hsapiens.UCSC.hg19.knownGene") for human genome hg38 and hg19, r Biocannopkg("TxDb.Mmusculus.UCSC.mm10.knownGene") and r Biocannopkg("TxDb.Mmusculus.UCSC.mm9.knownGene") for mouse genome mm10 and mm9, etc. User can also prepare their own TxDb object by retrieving information from UCSC Genome Bioinformatics and BioMart data resources by R function makeTxDbFromBiomart and makeTxDbFromUCSC. TxDb object should be passed for peak annotation.

All the peak information contained in peakfile will be retained in the output of annotatePeak. The position and strand information of nearest genes are reported. The distance from peak to the TSS of its nearest gene is also reported. The genomic region of the peak is reported in annotation column. Since some annotation may overlap, r Biocpkg("ChIPseeker") adopted the following priority in genomic annotation.

• Promoter
• 5' UTR
• 3' UTR
• Exon
• Intron
• Downstream
• Intergenic

Downstream is defined as the downstream of gene end.

r Biocpkg("ChIPseeker") also provides parameter genomicAnnotationPriority for user to prioritize this hierachy.

annotatePeak report detail information when the annotation is Exon or Intron, for instance "Exon (uc002sbe.3/9736, exon 69 of 80)", means that the peak is overlap with an Exon of transcript uc002sbe.3, and the corresponding Entrez gene ID is 9736 (Transcripts that belong to the same gene ID may differ in splice events), and this overlaped exon is the 69th exon of the 80 exons that this transcript uc002sbe.3 prossess.

Parameter annoDb is optional, if provided, extra columns including SYMBOL, GENENAME, ENSEMBL/ENTREZID will be added. The geneId column in annotation output will be consistent with the geneID in TxDb. If it is ENTREZID, ENSEMBL will be added if annoDb is provided, while if it is ENSEMBL ID, ENTREZID will be added.

Visualize Genomic Annotation

To annotate the location of a given peak in terms of genomic features, annotatePeak assigns peaks to genomic annotation in "annotation" column of the output, which includes whether a peak is in the TSS, Exon, 5' UTR, 3' UTR, Intronic or Intergenic. Many researchers are very interesting in these annotations. TSS region can be defined by user and annotatePeak output in details of which exon/intron of which genes as illustrated in previous section.

Pie and Bar plot are supported to visualize the genomic annotation.

plotAnnoPie(peakAnno)

plotAnnoBar(peakAnno)

Since some annotation overlap, user may interested to view the full annotation with their overlap, which can be partially resolved by vennpie function.

vennpie(peakAnno)

We extend r CRANpkg("UpSetR") to view full annotation overlap. User can user upsetplot function.

upsetplot(peakAnno)

We can combine vennpie with upsetplot by setting vennpie = TRUE.

upsetplot(peakAnno, vennpie=TRUE)

Visualize distribution of TF-binding loci relative to TSS

The distance from the peak (binding site) to the TSS of the nearest gene is calculated by annotatePeak and reported in the output. We provide plotDistToTSS to calculate the percentage of binding sites upstream and downstream from the TSS of the nearest genes, and visualize the distribution.

plotDistToTSS(peakAnno,
              title="Distribution of transcription factor-binding loci\nrelative to TSS")

Functional enrichment analysis

Once we have obtained the annotated nearest genes, we can perform functional enrichment analysis to identify predominant biological themes among these genes by incorporating biological knowledge provided by biological ontologies. For instance, Gene Ontology (GO)[@ashburner_gene_2000] annotates genes to biological processes, molecular functions, and cellular components in a directed acyclic graph structure, Kyoto Encyclopedia of Genes and Genomes (KEGG)[@kanehisa_kegg_2004] annotates genes to pathways, Disease Ontology (DO)[@schriml_disease_2011] annotates genes with human disease association, and Reactome[@croft_reactome_2013] annotates gene to pathways and reactions.

r Biocpkg("ChIPseeker") also provides a function, seq2gene, for linking genomc regions to genes in a many-to-many mapping. It consider host gene (exon/intron), promoter region and flanking gene from intergenic region that may under control via cis-regulation. This function is designed to link both coding and non-coding genomic regions to coding genes and facilitate functional analysis.

Enrichment analysis is a widely used approach to identify biological themes. I have developed several Bioconductor packages for investigating whether the number of selected genes associated with a particular biological term is larger than expected, including r Biocpkg("DOSE")[@yu_dose_2015] for Disease Ontology, r Biocpkg("ReactomePA") for reactome pathway, r Biocpkg("clusterProfiler")[@yu_clusterprofiler_2012] for Gene Ontology and KEGG enrichment analysis.

library(ReactomePA)

pathway1 <- enrichPathway(as.data.frame(peakAnno)$geneId)
head(pathway1, 2)

gene <- seq2gene(peak, tssRegion = c(-1000, 1000), flankDistance = 3000, TxDb=txdb)
pathway2 <- enrichPathway(gene)
head(pathway2, 2)
dotplot(pathway2)

More information can be found in the vignettes of Bioconductor packages r Biocpkg("DOSE")[@yu_dose_2015], r Biocpkg("ReactomePA"), r Biocpkg("clusterProfiler")[@yu_clusterprofiler_2012], which also provide several methods to visualize enrichment results. The r Biocpkg("clusterProfiler")[@yu_clusterprofiler_2012] is designed for comparing and visualizing functional profiles among gene clusters, and can directly applied to compare biological themes at GO, DO, KEGG, Reactome perspective.

ChIP peak data set comparison

Profile of several ChIP peak data binding to TSS region

Function plotAvgProf and tagHeatmap can accept a list of tagMatrix and visualize profile or heatmap among several ChIP experiments, while plotAvgProf2 and peakHeatmap can accept a list of bed files and perform the same task in one step.

Average profiles

## promoter <- getPromoters(TxDb=txdb, upstream=3000, downstream=3000)
## tagMatrixList <- lapply(files, getTagMatrix, windows=promoter)
##
## to speed up the compilation of this vigenette, we load a precaculated tagMatrixList
data("tagMatrixList")
plotAvgProf(tagMatrixList, xlim=c(-3000, 3000))

plotAvgProf(tagMatrixList, xlim=c(-3000, 3000), conf=0.95,resample=500, facet="row")

Peak heatmaps

tagHeatmap(tagMatrixList, xlim=c(-3000, 3000), color=NULL)

ChIP peak annotation comparision

The plotAnnoBar and plotDistToTSS can also accept input of a named list of annotated peaks (output of annotatePeak).

peakAnnoList <- lapply(files, annotatePeak, TxDb=txdb,
                       tssRegion=c(-3000, 3000), verbose=FALSE)

We can use plotAnnoBar to comparing their genomic annotation.

plotAnnoBar(peakAnnoList)

R function plotDistToTSS can use to comparing distance to TSS profiles among ChIPseq data.

plotDistToTSS(peakAnnoList)

Functional profiles comparison

As shown in section 4, the annotated genes can analyzed by r Biocpkg("clusterProfiler")[@yu_clusterprofiler_2012], r Biocpkg("DOSE")[@yu_dose_2015], r Biocpkg("meshes") and r Biocpkg("ReactomePA") for Gene Ontology, KEGG, Disease Ontology, MeSH and Reactome Pathway enrichment analysis.

The r Biocpkg("clusterProfiler")[@yu_clusterprofiler_2012] package provides compareCluster function for comparing biological themes among gene clusters, and can be easily adopted to compare different ChIP peak experiments.

genes = lapply(peakAnnoList, function(i) as.data.frame(i)$geneId)
names(genes) = sub("_", "\n", names(genes))
compKEGG <- compareCluster(geneCluster   = genes,
                         fun           = "enrichKEGG",
                         pvalueCutoff  = 0.05,
                         pAdjustMethod = "BH")
dotplot(compKEGG, showCategory = 15, title = "KEGG Pathway Enrichment Analysis")

Overlap of peaks and annotated genes

User may want to compare the overlap peaks of replicate experiments or from different experiments. r Biocpkg("ChIPseeker") provides peak2GRanges that can read peak file and stored in GRanges object. Several files can be read simultaneously using lapply, and then passed to vennplot to calculate their overlap and draw venn plot.

vennplot accept a list of object, can be a list of GRanges or a list of vector. Here, I will demonstrate using vennplot to visualize the overlap of the nearest genes stored in peakAnnoList.

genes= lapply(peakAnnoList, function(i) as.data.frame(i)$geneId)
vennplot(genes)

Statistical testing of ChIP seq overlap

Overlap is very important, if two ChIP experiment by two different proteins overlap in a large fraction of their peaks, they may cooperative in regulation. Calculating the overlap is only touch the surface. r Biocpkg("ChIPseeker") implemented statistical methods to measure the significance of the overlap.

Shuffle genome coordination

p <- GRanges(seqnames=c("chr1", "chr3"),
             ranges=IRanges(start=c(1, 100), end=c(50, 130)))
shuffle(p, TxDb=txdb)

We implement the shuffle function to randomly permute the genomic locations of ChIP peaks defined in a genome which stored in TxDb object.

Peak overlap enrichment analysis

With the ease of this shuffle method, we can generate thousands of random ChIP data and calculate the background null distribution of the overlap among ChIP data sets.

enrichPeakOverlap(queryPeak     = files[[5]],
                  targetPeak    = unlist(files[1:4]),
                  TxDb          = txdb,
                  pAdjustMethod = "BH",
                  nShuffle      = 50,
                  chainFile     = NULL,
                  verbose       = FALSE)

Parameter queryPeak is the query ChIP data, while targetPeak is bed file name or a vector of bed file names from comparison; nShuffle is the number to shuffle the peaks in targetPeak. To speed up the compilation of this vignettes, we only set nShuffle to 50 as an example for only demonstration. User should set the number to 1000 or above for more robust result. Parameter chainFile are chain file name for mapping the targetPeak to the genome version consistent with queryPeak when their genome version are different. This creat the possibility of comparison among different genome version and cross species.

In the output, qSample is the name of queryPeak and qLen is the the number of peaks in queryPeak. N_OL is the number of overlap between queryPeak and targetPeak.

Data Mining with ChIP seq data deposited in GEO

There are many ChIP seq data sets that have been published and deposited in GEO database. We can compare our own dataset to those deposited in GEO to search for significant overlap data. Significant overlap of ChIP seq data by different binding proteins may be used to infer cooperative regulation and thus can be used to generate hypotheses.

We collect about 17,000 bed files deposited in GEO, user can use getGEOspecies to get a summary based on speices.

GEO data collection

getGEOspecies()

The summary can also based on genome version as illustrated below:

getGEOgenomeVersion()

User can access the detail information by getGEOInfo, for each genome version.

hg19 <- getGEOInfo(genome="hg19", simplify=TRUE)
head(hg19)

If simplify is set to FALSE, extra information including source_name, extract_protocol, description, data_processing and submission_date will be incorporated.

Download GEO ChIP data sets

r Biocpkg("ChIPseeker") provide function downloadGEObedFiles to download all the bed files of a particular genome.

downloadGEObedFiles(genome="hg19", destDir="hg19")

Or a vector of GSM accession number by downloadGSMbedFiles.

gsm <- hg19$gsm[sample(nrow(hg19), 10)]
downloadGSMbedFiles(gsm, destDir="hg19")

Overlap significant testing

After download the bed files from GEO, we can pass them to enrichPeakOverlap for testing the significant of overlap. Parameter targetPeak can be the folder, e.g. hg19, that containing bed files. enrichPeakOverlap will parse the folder and compare all the bed files. It is possible to test the overlap with bed files that are mapping to different genome or different genome versions, enrichPeakOverlap provide a parameter chainFile that can pass a chain file and liftOver the targetPeak to the genome version consistent with queryPeak. Signifcant overlap can be use to generate hypothesis of cooperative regulation.By mining the data deposited in GEO, we can identify some putative complex or interacted regulators in gene expression regulation or chromsome remodelling for further validation.

Need helps?

If you have questions/issues, please visit ChIPseeker homepage first. Your problems are mostly documented. If you think you found a bug, please follow the guide and provide a reproducible example to be posted on github issue tracker. For questions, please post to Bioconductor support site and tag your post with ChIPseeker.

For Chinese user, you can follow me on WeChat (微信).

Session Information

Here is the output of sessionInfo() on the system on which this document was compiled:

sessionInfo()

References

ChIPseeker documentation built on March 6, 2021, 2 a.m.