plotClusterHeatmap {CATALYST}R Documentation

Plot cluster heatmap

Description

Plots heatmaps summarizing a clustering and/or metaclustering of interest.

Usage

plotClusterHeatmap(x, ...)

## S4 method for signature 'daFrame'
plotClusterHeatmap(x, hm2 = NULL, k = "meta20",
  m = NULL, fun = c("median", "mean"), cluster_anno = TRUE,
  split_by = NULL, scale = TRUE, draw_dend = TRUE,
  draw_freqs = FALSE, palette = rev(brewer.pal(11, "RdYlBu")))

Arguments

x

a daFrame.

...

optional arguments.

hm2

character string. Specifies the right-hand side heatmap. One of:

  • "abundances": cluster frequencies across samples

  • "state_markers": median cell state marker expressions across clusters (analogous to the left-hand side heatmap)

  • a character string/vector corresponding to one/multiple marker(s): median marker expressions across samples and clusters

k

character string. Specifies the clustering across which median marker expressions should be computed.

m

character string. Specifies the metaclustering to be shown. (This is for display only and will not effect any computations!)

fun

character string specifying the function to use as summary statistic.

cluster_anno

logical. Specifies if clusters should be annotated.

split_by

character string. Must corresponds to a column name of rowData(x). If specified, the data will be subset according to this variable, and multiple heatmaps will be drawn.

scale

logical. Specifies whether scaled values should be plotted. (see below for details)

draw_dend

logical. Specifies if the row dendrogram should be drawn.

draw_freqs

logical. Specifyies whether to display cell counts and proportions.

palette

character vector of colors to interpolate.

Details

Scaled values corresponds to cofactor arcsinh-transformed expression values scaled between 0 and 1 using 1 boundaries. Hierarchical clustering is performed on the unscaled data.

In its 1st panel, plotClusterHeatmap will display median (scaled, arcsinh-transformed) cell-type marker expressions (across all samples). Depending on argument hm2, the 2nd panel will contain one of:

Value

a HeatmapList-class object.

Author(s)

Helena Lucia Crowell helena.crowell@uzh.ch

References

Nowicka M, Krieg C, Weber LM et al. CyTOF workflow: Differential discovery in high-throughput high-dimensional cytometry datasets. F1000Research 2017, 6:748 (doi: 10.12688/f1000research.11622.1)

Examples

data(PBMC_fs, PBMC_panel, PBMC_md)
re <- daFrame(PBMC_fs, PBMC_panel, PBMC_md)

# run clustering
re <- cluster(re)

plotClusterHeatmap(re, hm2="abundances")
plotClusterHeatmap(re, hm2="abundances", draw_freqs=TRUE)
plotClusterHeatmap(re, hm2="state_markers", k="meta16", split_by='condition')
plotClusterHeatmap(re, hm2="pS6", k="meta12", m="meta8")
plotClusterHeatmap(re, hm2="abundances", scale=FALSE, draw_freqs=TRUE)


[Package CATALYST version 1.8.0 Index]