conversions: Convert between GRanges and PyRanges objects

In hpages/BiocPyRanges: Conveniently access PyRanges functionalities from Bioconductor

Convert between GRanges and PyRanges objects

Description

Two utility functions for converting back and forth between GRanges and PyRanges objects.

Usage

makePyRangesFromGRanges(gr)

makeGRangesFromPyRanges(pyr)

Arguments

gr	A GRanges object, or, more generally, it can be any GenomicRanges derivative.
pyr	A PyRanges object.

Value

makePyRangesFromGRanges returns a PyRanges object.

makeGRangesFromPyRanges returns a GRanges object.

References

PyRanges: https://github.com/biocore-NTNU/pyranges

PyRanges' documentation: https://biocore-NTNU.github.io/pyranges/

See Also

• GRanges objects defined in the GenomicRanges package.

• What is the BiocPyRanges package about.

Examples

## ---------------------------------------------------------------------
## A SIMPLE EXAMPLE
## ---------------------------------------------------------------------

gr <- GRanges("chr1", IRanges(1:3, 20))
pyr <- makePyRangesFromGRanges(gr)
pyr  # PyRanges objects follow the 0-based start convention!!

makeGRangesFromPyRanges(pyr)

## ---------------------------------------------------------------------
## TYPICAL BiocPyRanges WORKFLOW
## ---------------------------------------------------------------------

gr1 <- GRanges("chr1", IRanges(c(1:3, 15, 35, 51), c(20:22, 30, 50, 60)),
               strand=c("+", "+", "-", "+", "+", "+"))
gr1

## The typical BiocPyRanges workflow has 3 steps:

## STEP 1 - Go to PyRanges space (i.e. turn your GRange object into a
##          PyRanges object):
pyr1 <- makePyRangesFromGRanges(gr1)
pyr1

## STEP 2 - Operate in PyRanges space (i.e. perform some operation
##          provided by the PyRanges module on your PyRanges object):
pyr2 <- pyr1$merge()  # equivalent of GenomicRanges::reduce()

## STEP 3 - Come back to R space (i.e. turn result back into GRanges
##          object):
gr2 <- makeGRangesFromPyRanges(pyr2)

## Sanity check:
stopifnot(identical(reduce(gr1, ignore.strand=TRUE), gr2))

## ---------------------------------------------------------------------
## PERFORMANCE
## ---------------------------------------------------------------------

## PyRanges is very fast!
random_GRanges <- function(n, width) {
    gr_starts <- sample(50e6, n, replace=TRUE)
    chroms <- c("chr1", "chr2", "chr3")
    gr_chrom <- sample(factor(chroms, levels=chroms), n, replace=TRUE)
    gr_strand <- sample(levels(strand()), n, replace=TRUE)
    GRanges(gr_chrom, IRanges(gr_starts, width=width, names=seq_len(n)),
            strand=gr_strand)
}

set.seed(123)
gr1 <- random_GRanges(3e6, width=40)  # 3 million random ranges

system.time(gr2a <- reduce(gr1, ignore.strand=TRUE))

pyr1 <- makePyRangesFromGRanges(gr1)
system.time(pyr2 <- pyr1$merge())
gr2b <- makeGRangesFromPyRanges(pyr2)

## Sanity check:
stopifnot(identical(gr2a, gr2b))

hpages/BiocPyRanges documentation built on Aug. 21, 2023, 10:11 a.m.