Make Riverplot/Sankey diagram that shows label mapping across datasets

Creates a riverplot/Sankey diagram to show how independent cluster assignments from two datasets map onto a joint clustering. Prior knowledge of cell annotation for the given datasets is required to make sense from the visualization. Dataset original annotation can be added with the syntax shown in example code in this manual. The joint clustering could be generated with runCluster or set by any other metadata annotation.

Dataset original annotation can be inserted before running this function using cellMeta<- method. Please see example below.

This function depends on CRAN available package "sankey" and it has to be installed in order to make this function work.

Usage

plotSankey(
  object,
  cluster1,
  cluster2,
  clusterConsensus = NULL,
  minFrac = 0.01,
  minCell = 10,
  titles = NULL,
  prefixes = NULL,
  labelCex = 1,
  titleCex = 1.1,
  colorValues = scPalette,
  mar = c(2, 2, 4, 2)
)

Arguments

object

A liger object with all three clustering variables available.

cluster1, cluster2

Name of the variables in cellMeta(object) for the cluster assignments of dataset 1 and 2, respectively.

clusterConsensus

Name of the joint cluster variable to use. Default uses the default clustering of the object. Can select a variable name in cellMeta(object).

minFrac

Numeric. Minimum fraction of cluster for an edge to be shown. Default 0.05.

minCell

Numeric. Minimum number of cells for an edge to be shown. Default 10.

titles

Character vector of three. Customizes the column title text shown. Default uses the variable names cluster1, clusterConsensus and cluster2.

prefixes

Character vector of three. Cluster names have to be unique across all three variables, so this is provided to deduplicate the clusters by adding "prefixes[i]-" before the actual label. This will not be applied when no duplicate is found. Default NULL uses variable names. An NA value or a string with no character (i.e. "") does not add the prefix to the corresponding variable.

labelCex

Numeric. Amount by which node label text should be magnified relative to the default. Default 1.

titleCex

Numeric. Amount by which node label text should be magnified relative to the default. Default 1.1.

colorValues

Character vector of color codes to set color for each level in the consensus clustering. Default scPalette.

mar

Numeric vector of the form c(bottom, left, top, right) which gives the number of lines of margin to be specified on the four sides of the plot. Increasing the 2nd and 4th values can be helpful when cluster labels are long and extend out side of the plotting region. Default c(2, 2, 4, 2).

Value

No returned value. The sankey diagram will be displayed instead.

Note

This function works as a replacement of the function makeRiverplot in rliger <1.99. We decide to make a new function because the dependency adopted by the older version is archived on CRAN and will be no longer available.

Examples

# Make fake dataset specific labels from joint clustering result
cellMeta(pbmcPlot, "ctrl_cluster", "ctrl") <-
    cellMeta(pbmcPlot, "leiden_cluster", "ctrl")
cellMeta(pbmcPlot, "stim_cluster", "stim") <-
    cellMeta(pbmcPlot, "leiden_cluster", "stim")
if (requireNamespace("sankey", quietly = TRUE)) {
    plotSankey(pbmcPlot, "ctrl_cluster", "stim_cluster",
               titles = c("control", "LIGER", "stim"),
               prefixes = c("c", NA, "s"))
}