readMTX2IntSet: Create an InteractionSet from a BED file and Matrix Market...
In diffHic: Differential Analyis of Hi-C Data

Description Usage Arguments Details Value Author(s) See Also Examples

Read a set of Matrix Market Exchange Format files from disk and create an InteractionSet object.

1	readMTX2IntSet(mtx, bed, as.integer=TRUE)

`mtx`	A character vector containg paths to Matrix Market Exchange Format (MTX) files. Each file contains interaction read counts for one sample.
`bed`	String containing the path to the BED file specifying the genomic regions.
`as.integer`	Logical indicating whether the data should be read as integers. Otherwise values are stored in double-precision numbers.

Each MTX file is assumed to contain read counts for symmetric matrix representing the two-dimensional interaction space. Each row and column is assumed to correspond to contiguous bins of the genome, with coordinates specified by bed in standard BED format. This function will aggregate counts from all files to create an InteractionSet object that mimics the output of squareCounts.

The width value in the metadata of the output InteractionSet is set to the median width of the regions. The totals field in the output colData is also set to be equal to the sum of the counts in each MTX file. Note that these settings only make sense if the ContactMatrix objects cover binned regions.

This function can, in principle, read and merge any number of MTX files. However, for large data sets, consider reading each MTX file separately, subsetting it to interactions of interest and then creating the InteractionSet object. For example, subsetted contact matrices can be used to create an InteractionSet via mergeCMs.

An InteractionSet object containing interactions between regions from the BED file (in reverse-strict mode, see GInteractions). Each row corresponds to a unique interaction found in any of the MTX files, and contains the read counts across all files.

Gordon Smyth, with modifications by Aaron Lun

mergeCMs

library(Matrix)
tmp.loc <- tempfile()
dir.create(tmp.loc)

# Mocking up some MTX and BED files.
set.seed(110000)
A <- rsparsematrix(1000, 1000, density=0.1, symmetric=TRUE, 
    rand.x=function(n) round(runif(n, 1, 100)))
A.name <- file.path(tmp.loc, "A.mtx")
writeMM(file=A.name, A)

B <- rsparsematrix(1000, 1000, density=0.1, symmetric=TRUE, 
    rand.x=function(n) round(runif(n, 1, 100)))
B.name <- file.path(tmp.loc, "B.mtx")
writeMM(file=B.name, B)

GR <- GRanges(sample(c("chrA", "chrB", "chrC"), 1000, replace=TRUE),
    IRanges(start=round(runif(1000, 1, 10000)),
        width=round(runif(1000, 50, 500))))
GR <- sort(GR)
bed.name <- file.path(tmp.loc, "regions.bed")
rtracklayer::export.bed(GR, con=bed.name)

# Reading everything in.
iset <- readMTX2IntSet(c(A.name, B.name), bed.name)
iset