summarizeReads: Count reads lying within given regions.

Description

This function takes reads from e.g. ChIP-seq experiments and regions, e.g. promoters of genes, and assigns the number of overlapping reads to that region. This method was written particularly with regard to histone ChIP-seq experiments. Some histone modifications mainly occur at transcriptional start sites and thus ChIP-seq values can be assigned to genes by counting the number of reads within genes' pomoter regions. However, some genes may have several transcript and hence several promoters. Different options for handling multiple promoters are implemented. This method is also useful when integrating microarray expression data and ChIP-seq data, since most array platforms are gene centric and have probes that measure several transcripts.

Usage

summarizeReads(object, regions, summarize)

Arguments

object

A GRangesList with one GRanges object for each sample storing the ChIP-seq reads. The names of the GRangesList elements are used as sample names.

regions

An object of type GRangesList storing the promoter regions. Each element can be interpreted as gene or probe that has one or more promoters. The names of the lists' elements are used as row names. Alternatively, regions can be a GRanges object which as then handled like a GRangesList object with only one region in each list element. names of the GRanges are used as row names in this case.

summarize

Defines how regions with several ranges are handled. "average" means that the mean count of reads across all ranges is assigned to the region whereas "add" means that all counts are simply added (default).

Value

Returns a ChIPseqSet with number of rows equal to the length of regions and number of samples equal to the length of object.

Details

This function is usually applied after calling matchProbeToPromoter. When matchProbeToPromoter is used with mode "union", it is recommended to use "add". If the option "keepAll" had been used, one might want to use "average".

This method uses countOverlaps and counts each read that overlaps with at least one base.

Examples

Run this code

chipSeq <- GRangesList()
chipSeq[[1]] <- GRanges(seqnames=c("1", "1", "1", "1"),
    ranges=IRanges(start=c(97900, 198200, 198600, 202500),
        end=c(98100, 198400, 198800, 202700)),
    strand=c("+", "+", "+", "+"))
chipSeq[[2]] <- GRanges(seqnames=c("1", "1", "1", "1"),
    ranges=IRanges(start=c(97900, 198200, 198600, 300000),
        end=c(98100, 198400, 198800, 300200)),
    strand=c("+", "+", "+", "+"))
names(chipSeq) = c("sample1", "sample2")

promoters <- GRanges(seqnames=c("1", "1", "1"),
    ranges=IRanges(start=c(98000, 198000, 202000),
        end=c(101999, 201999, 205999)),
    strand=c("-", "+", "+"),
    probe=c("101", "102", "102"))
promoters <- split(promoters, elementMetadata(promoters)$probe)

chipSet <- summarizeReads(chipSeq, promoters, summarize="add")
chipVals(chipSet)