annotatePairs: Annotate bin pairs

Description

Annotate bin pairs based on features overlapping the anchor regions.

Usage

annotatePairs(data.list, regions, rnames=names(regions), indices, ...)

Arguments

data.list

An InteractionSet or a list of InteractionSet objects containing bin pairs.

regions

A GRanges object containing coordinates for the regions of interest.

rnames

A character vector containing names to be used for annotation.

indices

An integer vector or list of such vectors, indicating the cluster identity for each interaction in data.list.

...

Additional arguments to pass to findOverlaps.

Value

A list of two character vectors anchor1 and anchor2 is returned, containing comma-separated strings of names for entries in regions overlapped by the first and second anchor regions respectively. If indices is not specified, overlaps are identified to anchor regions of each interaction in data.list. Otherwise, overlaps are identified to anchor regions for any interaction in each cluster.

Details

Entries in regions are identified with any overlap to anchor regions for interactions in data.list. The names for these entries are concatenated into a comma-separated string for easy reporting. Typically, gene symbols are used in names, but other values can be supplied depending on the type of annotation. This is done separately for the first and second anchor regions so that potential interactions between features of interest can be identified.

If indices is supplied, all interactions corresponding to each unique index are considered to be part of a single cluster. Overlaps with all interactions in the cluster are subsequently concatenated into a single string. Cluster indices should range from [1, nclusters] for any given number of clusters. This means that the annotation for a cluster corresponding to a certain index can be obtained by subsetting the output vectors with that index. Otherwise, if indices is not set, all interactions are assumed to be their own cluster, i.e., annotation is returned for each interaction separately.

Multiple InteractionSet objects can be supplied in data.list, e.g., if the cluster consists of bin pairs of different sizes. This means that indices should also be a list of vectors where each vector indicates the cluster identity of the entries in the corresponding InteractionSet of data.list.

Examples

Run this code

# Setting up the objects.
a <- 10
b <- 20
cuts <- GRanges(rep(c("chrA", "chrB"), c(a, b)), IRanges(c(1:a, 1:b), c(1:a, 1:b)))
param <- pairParam(cuts)

all.combos <- combn(length(cuts), 2) 
y <- InteractionSet(matrix(0, ncol(all.combos), 1), 
    GInteractions(anchor1=all.combos[2,], anchor2=all.combos[1,], regions=cuts, mode="reverse"),
    colData=DataFrame(lib.size=1000), metadata=List(param=param, width=1))

regions <- GRanges(rep(c("chrA", "chrB"), c(3,2)), IRanges(c(1,5,8,3,3), c(1,5,8,3,4)))
names(regions) <- LETTERS[seq_along(regions)] 
out <- annotatePairs(y, regions=regions)

# Again, with indices:
indices <- sample(20, length(y), replace=TRUE)
out <- annotatePairs(y, regions=regions, indices=indices)

# Again, with multiple InteractionSet objects:
out <- annotatePairs(list(y, y[1:10,]), regions=regions, indices=list(indices, indices[1:10]))