R/MPCBS.EXTS.R

In HenrikBengtsson/aroma.core: Core Methods and Classes Used by 'aroma.*' Packages Part of the Aroma Framework

setMethodS3("drawCnRegions", "MPCBS", function(object, ...) {
  cnr <- extractCopyNumberRegions(object, ...)
  drawLevels(cnr, ...)
})


setMethodS3("extractCopyNumberRegions", "MPCBS", function(object, ...) {
  # To please R CMD check in case 'mpcbs' is not available
  # This might be need in order for it to work on CRAN. /HB 2010-01-04
  if (!isPackageInstalled("mpcbs")) {
    compute.var <- function(...) {}
    cross.platform.consensus <- function(...) {}
    # Not needed anymore
    rm(list=c("compute.var", "cross.platform.consensus"))
  }

  # Early error, iff package is not installed
  requireNamespace("mpcbs") || throw("Package not loaded: mpcbs")
  cross.platform.consensus <- mpcbs::cross.platform.consensus
  compute.var <- mpcbs::compute.var

  # According to example on help("mpcbs-package", package="mpcbs")
  # in mpcbs v1.0.0.

  fit <- object$fit
  chromosome <- object$chromosome
  K <- length(fit$yhat) # Number of sources

  # Compute platform specific variance terms.
  sigma2 <- double(K)
  for(kk in seq_len(K)) {
    y <- fit$y[[kk]]
    y <- y[is.finite(y)]
    sigma2[kk] <- compute.var(y)
  }

  # When data from multiple chromosomes are available,
  # put them in the list "fits".  Now, the list has length 1.
  fits <- vector("list", length=1)
  fits[[1]] <- fit
  platform.names <- seq_len(K)
  stdout <- capture.output({
    res <- cross.platform.consensus(fits, sigma2, platform.names, plots=FALSE)
  })

  # Target loci
  x <- fit$anchor$merged.pos

  # consensus.cn[t] is consensus copy number estimate at x[t]
  y <- res$consensus.cn[[1]]

  # Number of loci
  nbrOfLoci <- length(x)

  # Sanity check
  .stop_if_not(length(y) == nbrOfLoci)

  # Identify the locus indices where the regions starts and ends
  dy <- diff(y)
  starts <- c(1L, which(dy != 0)+1L)
  ends <- c(starts[-1]-1L, nbrOfLoci)

  # Sanity check
  .stop_if_not(length(starts) == length(ends))

  counts <- ends - starts + 1L

  # Get the mean levels of each region
  means <- y[starts]

  # Sanity check
  .stop_if_not(length(means) == length(starts))

  # Translate to genomic positions
  starts <- x[starts]
  ends <- x[ends]

  CopyNumberRegions(
    chromosome=chromosome,
    start=starts,
    stop=ends,
    mean=means,
    count=counts
  )
})