Skip to contents

Segment the Per-base Log2-FC of All Loci

Source: R/segmentationLFC.r
segmentationLFC.Rd

WARNING : coverage() must be run before segmentationLFC().

In brief, segmentationLFC() identifies segments with homogeneous Log2-FC along the loci of interest. segmentationLFC() can operate on each locus in parallel.

In details, for each locus, segmentationLFC() calls fpopw::Fpop_w() on the Log2-FC profile (possibly stranded) to retreive the segments boundaries (changepoints). The reference condition specified by the user is set to the denominator of the Log2-FC. The hyperparameter alpha specified by regions are returned as GRanges object.

Usage

segmentationLFC(
  data,
  modelSelectionType = "yao",
  outputDirectory = NULL,
  alpha = 2,
  segmentNeighborhood = FALSE,
  Kmax = NULL,
  alphas = NULL,
  gridSearch = FALSE,
  verbose = TRUE
)

Arguments

data

The List object returned by newExperiment().

modelSelectionType

A String. Select the penalty used by FPOP:

  • yao : Yao's penalty alpha*sigma^2*log(n).

outputDirectory

A String. Path to the output directory. On directory by locus will be create within outputDirectory.

alpha

A Double. Segmentation hyperparameter used in Yao's penalty: alpha*sigma^2*log(n). The number of changepoints returned by fpopw::Fpop_w() is a decreasing function of alpha.

segmentNeighborhood

A Logical. Indicate the weighted pDPA algorithm (Rigaill 2010 and 2015) has to be used to explore the space of segmentations.

Kmax

An Integer. Segmentations with 1 to Kmax segments are recovered using segment neighborhood.

alphas

A vector of Double. A series of alphas used by the grid search procedure.

gridSearch

A Logical. Indicate if a grid search has to be used to explore the space of segmentations.

verbose

A Logical. Should all the operations performed be displayed ?

Value

A GRanges object that contains the segment boundaries.