In LTLA/GenomicInteractions: Classes and Methods for Handling Genomic Interaction Data

require(knitr)
opts_chunk$set(error=FALSE, message=FALSE, warning=FALSE)
library(BiocStyle)
require(GenomicInteractions)

Overview

The r Biocpkg("GenomicInteractions") package offers a number of basic utility functions for working with GenomicInteractions objects. These will be described in this vignette.

Distance calculations

We are often interested in the distances between interacting regions on the linear genome, which can be used to determine if an interaction is local or distal. These distances are computed from a GenomicInteractions object with the pairdist method. To illustrate, let's construct some interactions involving multiple chromosomes:

all.regions <- GRanges(rep(c("chrA", "chrB"), c(10, 5)), 
    IRanges(c(0:9*10+1, 0:4*5+1), c(1:10*10, 1:5*5)))
index.1 <- as.integer(c(5, 15,  3, 12, 9, 10))
index.2 <- as.integer(c(1,  5, 11, 13, 7,  4))
gi <- GenomicInteractions(index.1, index.2, all.regions)
gi

By default, pairdist returns the distances between the midpoints of the anchor regions for each interaction. Any inter-chromosomal interactions will not have a defined distance along the linear genome, so a NA is returned instead.

pairdist(gi)

Different types of distances are obtained by specifying the type argument, e.g., "gap", "span", "diag". In addition, whether an interaction is intra-chromosomal or not can be determined with the intrachr function:

intrachr(gi)

Finding the bounding box

Multiple interactions are conveniently summarized by identifying the minimum bounding box. This refers to the smallest rectangle that can contain all interactions in the two-dimensional interaction space. We can then examine the coordinates of the bounding box rather than having to deal with each individual interaction separately. The code below computes the bounding box for all interactions within chromosome A:

all.regions <- GRanges("chrA", IRanges(1:10*10, width=10))
index.1 <- as.integer(c(1,8,10,2))
index.2 <- as.integer(c(9,2,1,5))
gi <- GenomicInteractions(index.1, index.2, all.regions)
gi
boundingBox(gi)

The boundingBox() function will automatically "reflect" all interactions so that they lie on the same side of the diagonal in the interaction space. This generally results in a smaller minimum bounding box, but assumes that the first and second anchors are interchangeable. If this is not the case, setting reflect=FALSE will not perform any reflection across the diagonal.

boundingBox(gi, reflect=FALSE)

Advanced users can also compute bounding boxes for groups of entries in gi. This is useful for summarizing clustered or overlapping interactions. (The default behaviour simply treats all input interactions as a single group.)

boundingBox(gi, c("X", "Y", "X", "Y"))

Note that the function will fail if any group contains both inter- and intra-chromosomal interactions.

all.regions2 <- GRanges("chrB", IRanges(1:10*10, width=10))
gi2 <- GenomicInteractions(index.1, index.2, all.regions2)
try(boundingBox(c(gi, gi2)))

Swapping anchors

The swapAnchors method ensures that the first anchor is always no greater than than the second anchor for each interaction. This eliminates redundant permutations of anchor regions and ensures that an interaction between regions #1 and #2 is treated the same as an interaction between regions #2 and #1. Obviously, this assumes that redundant permutations are uninteresting.

all.regions <- GRanges(rep(c("chrA", "chrB"), c(5, 5)), 
    IRanges(c(0:4*10+1, 0:4*5+1), c(1:5*10, 1:5*5)))
index.1 <- as.integer(c(5, 3, 3, 1, 9, 10))
index.2 <- as.integer(c(1, 5, 6, 9, 7,  4))
gi <- GenomicInteractions(index.1, index.2, all.regions)
gi
swapAnchors(gi)

Other options include reversing the swapping order, such that the second anchor is always not less than the first anchor:

swapAnchors(gi, mode="reverse")

... or simply swapping all anchors from the original order:

swapAnchors(gi, mode="all")

Finding distal anchors

Given a "viewpoint" anchor region, the findDistalAnchors() function will identify all anchor regions that interact with any interval within the viewpoint via the interactions in a GenomicInteractions object. To illustrate, consider the following GenomicInteractions object:

all.regions <- GRanges("chrA", IRanges(1:10*10, width=10))
index.1 <- as.integer(c(1, 8, 10, 2, 6, 2, 1))
index.2 <- as.integer(c(9, 5,  4, 5, 4, 8, 2))
gi <- GenomicInteractions(index.1, index.2, all.regions)
gi

Let r y <- GRanges("chrA:1-25") be the viewpoint of interest. It is straightforward to identify all anchor regions that interact with y via gi:

findDistalAnchors(gi, y)

If both anchors of an interaction overlap a viewpoint, both of them are retained by default. These can be discarded with local=FALSE for a stricter definition of distal interactors:

findDistalAnchors(gi, y, local=FALSE)

One can view this function as taking a "cross-section" of the interaction space at the specified viewpoint, thus projecting all the interactions involving the viewpoint onto the linear genome. This mimics what is done in 4C experiments and may be more convenient for interpretation if only one viewpoint is of interest.

Wrapping up

Details for any given method can be found through the standard documentation, e.g., ?boundingBox to get the man page for the boundingBox() method. Further questions on r Biocpkg("GenomicInteractions") functionality should be asked on the Bioconductor support site.

sessionInfo()

LTLA/GenomicInteractions documentation built on June 24, 2019, 2:09 p.m.