outputScaleSpaceTree: Output list of all features for a given scale-space map

In walter-ca/Scale4C: Scale4C: an R/Bioconductor package for scale-space transformation of 4C-seq data

Description

This function provides a list of features for a given fingerprint map in scale-space, with position and range of sigma for which the feature in question exists

Usage

outputScaleSpaceTree(data, outputPeaks = TRUE, useLog = TRUE, 
    useIndex = TRUE)

Arguments

data	a Scale4C object with singularity data
outputPeaks	If TRUE, output GRanges peak list only, if FALSE, also output valley data in a larger table
useLog	If TRUE, use a log2 transformation on the square sigma values (fewer changes and fewer singularities for high sigma, in contrast to low sigma)
useIndex	If TRUE, use fragment position

Details

Similar to plotTesselation, outputScaleSpaceTree analyzes a list of singular points and calculates corresponding features, i.e. 'peaks' and 'valleys'. Each singular point marks the disappearance (or occurrence, depending on the view) of a feature in scale space: With increasing square sigma as smoothing parameter for the Gauss kernel, smaller features are merged into larger features. In case of Gauss smoothing, one feature is always surrounded by two features of the opposite type, e.g. a 'peak' is surrounded by two 'valleys'. If a 'peak' is smoothed out, it is replaced by a new valley formed of the former peak's adjacent valleys. The singularity list contains only direct information on those 'central' features; outputScaleSpaceTree adds data on the direct neighbours / adjacent features and also provides the sigma ranges for the features as a measure of their stability throughout the smoothing process. Mean read counts for the identitied features are also provided ("signal"). If outputPeaks is true, a reduced list of peaks is printed, while omitting valleys or the central-left-right structural information.

Value

A GRanges object that includes all features as identified through singular points, plus 'neighbour features' at each side (each 'peak' is surrounded by two 'valleys' and vice versa for Gauss kernel smoothing), with positions and range of sigma for which the feature in question remains stable

Author(s)

Carolin Walter

Examples

    # read prepared example data
    data(liverDataVP)
    output = outputScaleSpaceTree(liverDataVP, useLog = FALSE)
    head(output) 

walter-ca/Scale4C documentation built on May 5, 2019, 9:03 p.m.