Plot spatial domain clustering benchmarks

Visualize a spatial_benchmark_result from RunSpatialBenchmark(), or a summary data.frame. Spatial results default to a publication-oriented overview that pairs clustering quality with runtime and peak-memory efficiency. For scRNA integration metrics such as LISI, use IntegrationBenchmarkPlot().

Usage

SpatialBenchmarkPlot(
  srt = NULL,
  data = NULL,
  features = NULL,
  metrics = NULL,
  tool_name = NULL,
  reduction = NULL,
  plot_type = c("auto", "overview", "quality", "efficiency", "heatmap", "feature",
    "boxplot", "bar", "funkyheatmap"),
  sort_by = c("quality", "method", "runtime", "memory"),
  show_values = TRUE,
  show_status = TRUE,
  resource_scale = c("auto", "linear", "log10"),
  plot_boxplot = TRUE,
  boxplot_jitter = FALSE,
  combine = TRUE,
  nrow = NULL,
  ncol = NULL,
  byrow = TRUE,
  pt.size = NULL,
  pt.alpha = 1,
  palette = "Chinese",
  palcolor = NULL,
  theme_use = "theme_scop",
  theme_args = list(),
  verbose = TRUE,
  ...
)

Arguments

srt: A Seurat object.
data: Optional spatial_benchmark_result / benchmark_result object or summary data.frame containing at least metric and value, and optionally method, workflow, and direction.
features: Metadata columns containing per-cell benchmark scores.
metrics: One or more summary metric names to visualize. Default is NULL, which uses all available summary metrics.
tool_name: Tool entries created by benchmark-related workflows. This can be a character vector. For per-cell metrics, benchmark columns are resolved from tool entries that contain colnames; for summary metrics, entries containing summary or metrics$summary are used.
reduction: Dimensional reduction used for per-cell feature plots. Default is NULL, which uses the reduction stored in tool_name when available, otherwise DefaultReduction().
plot_type: Plot type. "overview", "quality", "efficiency", and "heatmap" consume spatial benchmark results. Existing "feature", "boxplot", "bar", and "funkyheatmap" modes remain supported.
sort_by: Method ordering for spatial benchmark plots. "quality" sorts by the mean selected quality metric; other choices sort by method, runtime, or peak memory.
show_values: Whether to print raw metric values on quality and heatmap panels.
show_status: Whether the overview should add a status strip for failed, unavailable, or timed-out methods.
resource_scale: Resource-axis transformation. "auto" independently uses log10 for runtime or memory when positive values span at least tenfold.
plot_boxplot: Whether to add the summary boxplot when per-cell metrics are shown.
boxplot_jitter: Whether to overlay jittered points on the boxplot.
combine, nrow, ncol, byrow: Combine plots with patchwork. combine = FALSE returns a list of ggplots.
pt.size, pt.alpha: Point size and transparency. pt.size = NULL scales with sqrt(n) (minimum 0.3). Rasterized points keep at least a two-pixel radius at raster.dpi = c(512, 512) and scale with raster.dpi.
palette, palcolor: Palette name (thisplot::show_palettes) or custom colors.
theme_use, theme_args: Theme name or function, plus extra theme arguments.
verbose: Whether to print the message. Default is TRUE.
...: The message to print.

Value

A ggplot, patchwork plot, or funkyheatmap object depending on the selected mode. If combine = FALSE in per-cell mode, a named list of plots is returned.

Examples

metrics_df <- data.frame(
  method = c("Raw", "Raw", "Harmony", "Harmony"),
  metric = c("batch_ASW_mixing", "celltype_ASW", "batch_ASW_mixing", "celltype_ASW"),
  value = c(0.42, 0.71, 0.68, 0.66)
)
SpatialBenchmarkPlot(
  data = metrics_df,
  plot_type = "bar"
)


data("pbmcmultiome_sub", package = "scop")
pbmcmultiome_sub[["MethodA_batch_LISI"]] <-
  seq_len(ncol(pbmcmultiome_sub)) / ncol(pbmcmultiome_sub)
pbmcmultiome_sub[["MethodB_batch_LISI"]] <-
  rev(pbmcmultiome_sub[["MethodA_batch_LISI", drop = TRUE]])
SpatialBenchmarkPlot(
  pbmcmultiome_sub,
  features = c("MethodA_batch_LISI", "MethodB_batch_LISI"),
  plot_type = "boxplot"
)

Usage

Arguments

Value

See also

Examples