Skip to contents

Plot a circular cell annotation tree

Source: R/CellTreePlot.R
CellTreePlot.Rd

Show cell proportions and optional numeric rings. Requires ggforce.

Usage

CellTreePlot(
  object,
  levels,
  count = NULL,
  tracks = NULL,
  palette = "Dark2",
  palcolor = NULL,
  label.size = 3,
  show_proportion = TRUE,
  title = NULL,
  return_data = FALSE,
  show_internal_labels = FALSE
)

Arguments

object

A Seurat object or a data frame containing cell annotations.

levels

One or more unique annotation column names, broadest first. Labels must be nonmissing and nonblank. The names .count and .proportion are reserved.

count

Column of nonnegative whole counts with a finite positive total. With NULL, each row counts as one cell.

tracks

Data frame keyed by unique full annotation paths in object, with one numeric column per ring (inner to outer). Rings share a zero-centered scale. Missing values and omitted paths appear gray.

palette, palcolor

Palette (default "Dark2") and optional custom colors. Named colors override matching broad groups; other groups use the palette. Unnamed colors follow group order.

label.size

Positive, finite leaf label size (default 3).

show_proportion

Logical. Label each leaf with its share of the total count (default TRUE).

title

Plot title (default NULL, no title).

return_data

Logical. Include the leaf and track data (default FALSE).

show_internal_labels

Logical. Label intermediate levels (default FALSE).

Value

A ggplot object that accepts layers with +. With return_data = TRUE, a list with plot, leaves (one row per annotation path with .count and .proportion), and tracks (aligned to the leaves, or NULL).

References

https://doi.org/10.1126/science.adt3130

https://mp.weixin.qq.com/s/F7cgHFmpEAhwhrzAN0NvUA

See also

ClusterTreePlot(), CellStatPlot()

Examples

data(pancreas_sub)
hierarchy <- c("CellType", "SubCellType")
CellTreePlot(pancreas_sub, hierarchy)


meta <- pancreas_sub[[]]
scores <- stats::aggregate(meta[c("S_score", "G2M_score")], meta[hierarchy], mean)
CellTreePlot(pancreas_sub, hierarchy, tracks = scores)


data(pbmcmultiome_sub)
pbmc <- pbmcmultiome_sub[[]]
groups <- c(
  B = "B cells", "CD4 T" = "T cells", "CD8 T" = "T cells",
  NK = "NK cells", DC = "Myeloid", Mono = "Myeloid"
)
pbmc$Group <- unname(groups[as.character(pbmc$CellType)])
hierarchy <- c("Group", "CellType", "SubCellType")

metrics <- c("nCount_RNA", "nFeature_RNA", "nCount_ATAC", "nFeature_ATAC")
tracks <- stats::aggregate(pbmc[metrics], pbmc[hierarchy], stats::median)
subset_medians <- vapply(pbmc[metrics], stats::median, numeric(1))
tracks[metrics] <- Map(
  function(x, center) log2(x / center),
  tracks[metrics], subset_medians
)
names(tracks)[match(metrics, names(tracks))] <- c(
  "RNA counts", "RNA genes", "ATAC counts", "ATAC peaks"
)

CellTreePlot(pbmc, hierarchy, tracks = tracks, label.size = 2.25)


genes <- c("MS4A1", "CD3D", "NKG7", "LST1")
rna <- SeuratObject::LayerData(
  pbmcmultiome_sub,
  assay = "RNA", layer = "counts", features = genes
)
detected <- as.data.frame(t(as.matrix(rna[genes, rownames(pbmc)] > 0)))
gene_tracks <- stats::aggregate(detected, pbmc[hierarchy], mean)
subset_detection <- colMeans(detected)
gene_tracks[genes] <- Map(
  function(x, center) 100 * (x - center),
  gene_tracks[genes], subset_detection
)
CellTreePlot(pbmc, hierarchy, tracks = gene_tracks, label.size = 2.25)