CNPBayes (version 1.2.2)

consensusCNP: Identify consensus start and stop coordinates of a copy number polymorphism

Description

The collection of copy number variants (CNVs) identified in a study can be encapulated in a GRangesList, where each element is a GRanges of the CNVs identified for an individual. (For a study with 1000 subjects, the GRangesList object would have length 1000 if each individual had 1 or more CNVs.) For regions in which CNVs occur in more than 2 percent of study participants, the start and end boundaries of the CNVs may differ because of biological differences in the CNV size as well as due to technical noise of the assay and the uncertainty of the breakpoints identified by a segmentation of the genomic data. Among subjects with a CNV called at a given locus, the consensusCNP function identifies the largest region that is copy number variant in half of these subjects.

Usage

consensusCNP(grl, transcripts, min.width = 2000, max.width = 2e+05, min.prevalance = 0.02)

Arguments

grl

A GRangesList of all CNVs in a study -- each element is the collection of CNVs for one individual.

transcripts

a GRanges object containing annotation of genes or transcripts (optional)

min.width

length-one integer vector specifying the minimum width of CNVs

max.width

length-one integer vector specifying the maximum width of CNVs

min.prevalance

a length-one numeric vector specifying the minimum prevalance of a copy number polymorphism. Must be in the interval [0,1]. If less that 0, this function will return all CNV loci regardless of prevalance. If greater than 1, this function will return a length-zero GRanges object

Value

a GRanges object providing the intervals of all identified CNPs above a user-specified prevalance cutoff.

Examples

library(GenomicRanges)
##
## Simulate 2 loci at which CNVs are common
##
set.seed(100)
starts <- rpois(1000, 100) + 10e6L
ends <- rpois(1000, 100) + 10.1e6L
cnv1 <- GRanges("chr1", IRanges(starts, ends))
cnv1$id <- paste0("sample", seq_along(cnv1))

starts <- rpois(500, 1000) + 101e6L
ends <- rpois(500, 1000) + 101.4e6L
cnv2 <- GRanges("chr5", IRanges(starts, ends))
cnv2$id <- paste0("sample", seq_along(cnv2))

##
## Simulate a few other CNVs that are less common because they are
## very large, or because they occur in regions that in which copy
## number alerations are not common
##
cnv3 <- GRanges("chr1", IRanges(9e6L, 15e6L), id="sample1400")
starts <- seq(5e6L, 200e6L, 10e6L)
ends <- starts + rpois(length(starts), 25e3L)
cnv4 <- GRanges("chr1", IRanges(starts, ends),
                id=paste0("sample", sample(1000:1500, length(starts))))

all_cnvs <- suppressWarnings(c(cnv1, cnv2, cnv3, cnv4))
grl <- split(all_cnvs, all_cnvs$id)
cnps <- consensusCNP(grl)

##
## 2nd CNP is filtered because of its size
##
truth <- GRanges("chr1", IRanges(10000100L, 10100100L))
seqinfo(truth) <- seqinfo(grl)
identical(cnps, truth)

##
## Both CNVs identified
##
cnps <- consensusCNP(grl, max.width=500e3)
truth <- GRanges(c("chr1", "chr5"),
                 IRanges(c(10000100L, 101000999L),
                         c(10100100L, 101400999L)))
seqlevels(truth, force=TRUE) <- seqlevels(grl)
seqinfo(truth) <- seqinfo(grl)
identical(cnps, truth)