Atomic clustree plot — ClustreePlotAtomic • plotthis

Core implementation for clustree (clustering tree) plots. This internal function renders a tree-based visualisation showing how cluster assignments change across increasing clustering resolutions. It is called by the exported ClustreePlot() function.

Usage

ClustreePlotAtomic(
  data,
  prefix,
  flip = FALSE,
  alpha = 0.85,
  palette = "Paired",
  palcolor = NULL,
  palreverse = FALSE,
  edge_palette = "Spectral",
  edge_palcolor = NULL,
  aspect.ratio = 1,
  legend.position = "right",
  legend.direction = "vertical",
  title = NULL,
  subtitle = NULL,
  xlab = NULL,
  ylab = NULL,
  expand = c(0.1, 0.1),
  theme = "theme_this",
  theme_args = list(),
  ...
)

Arguments

data: A data frame.
prefix: A character string specifying the common prefix of the resolution columns in data. All columns whose names start with this prefix are selected as resolution columns. The suffix after the prefix is parsed as a numeric resolution value. Supports "_" and "." as separators between the prefix and the resolution value (e.g. "res_0.5" or "p.0.5").
flip: A logical value. If TRUE, the tree is flipped so that resolutions are displayed on the x-axis (left to right) and cluster assignments are shown as row labels on the y-axis. Default: FALSE.
alpha: A numeric value in [0, 1] specifying the transparency of the nodes in the clustree plot. Only used when node_alpha is not explicitly provided via .... Default: 0.85.
palette: A character string specifying the palette to use. A named list or vector can be used to specify the palettes for different split_by values.
palcolor: A character string specifying the color to use in the palette. A named list can be used to specify the colors for different split_by values. If some values are missing, the values from the palette will be used (palcolor will be NULL for those values).
palreverse: A logical value indicating whether to reverse the palette. Default is FALSE.
edge_palette: A character string specifying the palette name for the edge colour gradient. Edges are coloured by the number of transitioning cells between clusters at adjacent resolutions, using ggraph::scale_edge_color_gradientn(). Default: "Spectral".
edge_palcolor: A character vector of custom colours for the edge colour gradient. When NULL (the default), colours are derived from edge_palette.
aspect.ratio: A numeric value specifying the aspect ratio of the plot.
legend.position: A character string specifying the position of the legend. if waiver(), for single groups, the legend will be "none", otherwise "right".
legend.direction: A character string specifying the direction of the legend.
title: A character string specifying the title of the plot. A function can be used to generate the title based on the default title. This is useful when split_by is used and the title needs to be dynamic.
subtitle: A character string specifying the subtitle of the plot.
xlab: A character string specifying the x-axis label.
ylab: A character string specifying the y-axis label.
expand: The values to expand the x and y axes. It is like CSS padding. When a single value is provided, it is used for both axes on both sides. When two values are provided, the first value is used for the top/bottom side and the second value is used for the left/right side. When three values are provided, the first value is used for the top side, the second value is used for the left/right side, and the third value is used for the bottom side. When four values are provided, the values are used for the top, right, bottom, and left sides, respectively. You can also use a named vector to specify the values for each side. When the axis is discrete, the values will be applied as 'add' to the 'expansion' function. When the axis is continuous, the values will be applied as 'mult' to the 'expansion' function. See also https://ggplot2.tidyverse.org/reference/expansion.html
theme: A character string or a theme class (i.e. ggplot2::theme_classic) specifying the theme to use. Default is "theme_this".
theme_args: A list of arguments to pass to the theme function.
...: Additional arguments passed to clustree::clustree(). Commonly used overrides include node_size_range, node_text_size, layout (default: "sugiyama"), show_axis, and node_text_colour. Note that x (the data) and prefix are set internally and cannot be overridden here.

Value

A ggplot object with height and width attributes.

Details

The function expects a data frame containing columns named with a common prefix followed by numeric resolution values (e.g. "res_0.1", "res_0.3", "res_0.5"). Each column represents cluster assignments at a given resolution, and the tree visualises how cells (rows) transition between clusters as resolution increases.

The core layout is produced via clustree::clustree(), which computes a ggraph layout (default: "sugiyama") linking clusters across adjacent resolutions. Nodes represent clusters and edges represent cell transitions; edge width and colour encode the number of transitioning cells. Cluster labels are overlaid on each node via ggplot2::geom_text().

Key features:

Resolution-level node colouring — each resolution level receives a distinct colour from palette via ggplot2::scale_color_manual().
Edge gradient — edges are coloured by transition count using ggraph::scale_edge_color_gradientn() with the edge_palette gradient.
Flip support — flip = TRUE places resolutions on the x-axis with cluster IDs as y-axis row labels for left-to-right reading.
Automatic dimension calculation — plot height and width are sized based on the number of resolutions, number of clusters, and legend configuration, with different formulae for flipped vs. non-flipped layouts.

Architecture

ClustreePlotAtomic executes the following steps:

Column selection — selects all columns whose names start with prefix via dplyr::starts_with().
NA filtering — removes rows with any missing values via stats::complete.cases().
Expansion normalisation — norm_expansion() parses the expand argument for continuous x and y axes.
Empty data guard — stops with an informative error if no columns match the prefix or no complete rows remain.
Resolution parsing — extracts the suffix after prefix from each column name and strips leading separators ("_" or ".") if present. Values are converted to numeric and sorted. Column names are rewritten to use the original prefix for consistent clustree processing.
Cluster counting — computes the number of unique clusters at the highest resolution (max_clusters), used for dimension calculation in the non-flipped layout.
Argument merging — builds a call to clustree::clustree() by merging user-supplied ... with sensible defaults for node_alpha, edge_width, show_axis, layout ("sugiyama"), node_size_range (c(6, 12)), node_text_size (3), and node_text_colour ("black").
clustree execution — calls gglogger::register(clustree::clustree) with the merged arguments to produce the base ggplot object.
Node labels — overlays ggplot2::geom_text() with the cluster label from the clustree output, using the configured text size and colour.
Resolution colour scale — applies ggplot2::scale_color_manual() with one colour per resolution level using palette_this(). The colour legend is suppressed (the edge gradient legend serves as the primary guide).
Edge colour scale — applies ggraph::scale_edge_color_gradientn() with the edge_palette palette, encoding transition counts via a colour-bar guide with black frame and ticks.
Theme and labels — applies the selected theme and sets title, subtitle, x-axis label, and y-axis label (default y-label is the prefix with trailing separator stripped).
Flip branch (flip = TRUE):
- Resolutions are placed on the x-axis (horizontal); clusters become y-axis rows.
- Width scales with max(3, 0.5 + nres * 1.0); height scales with max(3, 0.5 + max_clusters * 0.5) (or via aspect.ratio).
- scale_y_reverse() + coord_flip() reorients the layout so resolutions read left-to-right.
- Horizontal grid lines are dashed grey80; y-axis text and ticks are suppressed.
Non-flip branch (flip = FALSE, default):
- Resolutions are on the y-axis (rows); clusters spread across the x-axis (columns).
- Width scales with max(3, 0.5 + max_clusters * 0.5); height scales with max(3, 0.5 + nres * 1.0) (or via aspect.ratio).
- coord_cartesian(clip = "off") and scale_y_continuous() show resolution labels on the y-axis.
- Vertical grid lines are dashed grey80; x-axis text and ticks are suppressed.
Dimension storage — calculate_plot_dimensions() computes height and width attributes, stored on the ggplot object.