CompareOnBranches: Compare feature values over different branches

Description Usage Arguments

View source: R/pseudotime.R

Description

Compare feature values over different branches

Usage

1
2
3
4
5
6
7
CompareOnBranches(X, Paths, TargetPG, BootPG = NULL,
  OnlyIsomorphicBoot = TRUE, GroupsLab = NULL, PlotOrg = TRUE,
  TrajCol = FALSE, Conf = 0.95, AllBP = FALSE, TrimmingRadius = Inf,
  Partition, PrjStr, Main = "", ylab = "Feature value", alpha = 0.3,
  ScalePT = FALSE, Features = 4, facet_rows = 3, facet_cols = 3,
  Mode = "Var", ModePar = NULL, Log = FALSE, Span = 0.1, n.cores = 1,
  ClusType = "SOCK", ReturnGenes = FALSE)

Arguments

X

the data matrix, note that features must be names, i.e. column names must be non-empty

Paths

a list of paths used to construct the pseudotime. Note that the beginning of the pseudotime will be placed in correspondence of the fisrst node for each path. Hence the comparison is meaningfull only if all the path begin from the same node (Although, this will not be enforced by the function)

TargetPG

the ElPiGraph structure to be used

BootPG

if not NULL, a list of resampled ElPiGraph structures that will be used to compute confidence intervals

OnlyIsomorphicBoot

boolean. If TRUE only reampled ElPiGraph structures with the same number of branching points of the target structure will be used to compute confidence intervals. If FALSE, all of the structures will be used.

GroupsLab

character of factor, category information for the cells present in the matrix.

PlotOrg

boolean, should the original gene expression be reported. If false oly the smooth will be plotted.

TrajCol

boolean, should the points be colored by trajectory? If FALSE GroupsLab will be used.

Conf

the confidence level.

AllBP

boolean. If TRUE, the confidence level be displayed for all the branching points present on the selected branches. If FALSE, the confidence level will be displayed only for the branching points

TrimmingRadius

numeric, the trimming radius used to project the position of the branching points of the resampled trees on the pseudotime

Partition

A partition vactor attributing each point to at most one node

PrjStr

A projecturion structure, as produced by project_point_onto_graph

Main

string, the main title of the plot

ylab

string, the label of the y axis

alpha

numeric between 0 and 1, the trasparency of the points

ScalePT

boolean, should the pseudotime be normalized across the branches?

Features

either a string vector or a positive integer. If a string vector, it will be interpreted as the name of the features to plot (matched to the colnames of X). If positive integer, it will be interpreted as the number of the topmost interesting features according to the metric specified by Mode

facet_rows

integer, the number of rows per panel

facet_cols

integer, the number of cols per panel

Mode

string, the feature seletion mode used to determine the most interesting genes. It can be

  • 'Var' The genes with the largest variance across the points associated to the paths

  • 'Fano' The genes with the largest Fano factor (variance over mean) across the points associated to the paths

  • 'Mad' The genes with the largest median absolute deviation across the points associated to the paths

  • 'Mad/Median' The genes with the largest ratio of median absolute deviation to median across the points associated to the paths

  • 'MI' The genes with the largest mutual information (across all the paths)

  • 'MI.DB' The genes with the largest mutual inforlation (across differential paths)

  • 'KW' The genes with the largest difference, as computed by the Kruskal-Wallis test (across differential paths)

  • 'Cor.DB' The genes whose product of Spearman correlation (w.r.t. the pseudotime ordering) is the largest

ModePar

additional parameters for the feature selection mode used

Log

boolean, should a log scale be used on the y axis

Span

the span parameter of the loess fitter

n.cores

integer, the number of parallel processes to use.

ClusType

string, the type of cluster to use if n.cores is larger than 1

ReturnGenes

boolean, should the list of genes be returned instead of the plots?


Albluca/ElPiGraph.R documentation built on June 10, 2018, 5:22 p.m.