diffHic (version 1.4.2)

domainDirections: Calculate domain directionality

Description

Collect directionality statistics for domain identification with genomic bins.

Usage

domainDirections(files, param, width=50000, span=10)

Arguments

files

a character vector containing paths to the index files generated from each Hi-C library

param

a pairParam object containing read extraction parameters

width

an integer scalar specifying the width of each bin in base pairs

span

an integer scalar specifying the distance to consider for up/downstream interactions

Value

A GRanges object with one entry for each bin in the genome. The Up and Down fields in the metadata contain the total average count for the up-/downstream interactions involving each bin.

Details

The genome is partitioned into bins of size width. For each bin, this function computes the total count for all bin pairs involving that bin and the span upstream bins. This is repeated for the span downstream bins. If multiple files are specified, the total of the averages across all libraries for each bin pair is computed instead.

The total up- and downstream counts are returned in the Up and Down fields of the output GRanges object. This can be used to compute a directionality statistic, e.g., as defined by Dixon et al, or by computing the log-fold change between fields. The latter is easier to compute and model with a Gaussian distribution. In any case, the directionalities can be used in a HMM to identify domains - see the user's guide for more details.

References

Dixon JR et al. (2012). Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature 485:376-380.

See Also

squareCounts

Examples

hic.file <- system.file("exdata", "hic_sort.bam", package="diffHic")
cuts <- readRDS(system.file("exdata", "cuts.rds", package="diffHic"))
param <- pairParam(fragments=cuts)

# Setting up the parameters
fout <- "output.h5"
invisible(preparePairs(hic.file, param, file=fout))

# Not really that informative; see user's guide.
out <- domainDirections(fout, param, width=10)
out
domainDirections(fout, param, width=10, span=1)

# Calculating directionality log-FC with a large prior.
dir.logFC <- log2((out$Up+10)/(out$Down+10))
dir.logFC

# Calculating directionality index with Dixon's method.
dixon.stat <- sign(out$Up-out$Down)*2*((out$Up-out$Down)/(out$Up+out$Down))^2
dixon.stat