Prioritize perturbed cell types using Augur

Usage

RunAugur(
  srt,
  celltype.by,
  label.by,
  assay = NULL,
  layer = "counts",
  backend = c("cpp", "r"),
  features = NULL,
  n_subsamples = 50,
  subsample_size = 20,
  folds = 3,
  min_cells = NULL,
  var_quantile = 0.5,
  feature_perc = 0.5,
  cores = 1,
  select_var = TRUE,
  augur_mode = c("default", "velocity", "permute"),
  classifier = c("rf", "lr"),
  rf_params = list(trees = 100, mtry = 2, min_n = NULL, importance = "accuracy"),
  lr_params = list(mixture = 1, penalty = "auto"),
  prefix = "augur",
  tool_name = "Augur",
  add_meta = TRUE,
  verbose = TRUE,
  ...
)

Arguments

srt: A Seurat object.
celltype.by: Metadata column or vector defining the cell types used by Augur.
label.by: Metadata column or vector defining the labels to predict, such as condition, stimulation, sample type, or technology.
assay: Assay used to extract the feature matrix. If NULL, the default assay is used.
layer: Assay layer used as the feature matrix.
backend: Backend used to run Augur. "cpp" uses a parity-preserving scop R/C++ backend that keeps Augur's feature selection, sampling, random forest, and metric semantics without requiring the Augur package. "r" calls the native Augur::calculate_auc implementation.
features: Features used by Augur. If NULL, all features in assay are used.
n_subsamples, subsample_size, folds, min_cells, var_quantile, feature_perc, select_var, augur_mode, classifier, rf_params, lr_params: Arguments passed to Augur::calculate_auc.
cores: Number of cores used by Augur.
prefix: Prefix for metadata columns written to srt@meta.data.
tool_name: Name of the srt@tools entry used to store Augur results.
add_meta: Whether to write prefix_auc and prefix_rank metadata columns back to each cell by matching celltype.by.
verbose: Whether to print the message. Default is TRUE.
...: Additional arguments passed to Augur::calculate_auc.

Value

A Seurat object with native Augur results stored in srt@tools[[tool_name]]. When add_meta = TRUE, cell-level metadata columns prefix_auc and prefix_rank are added for use with existing scop plotting functions such as FeatureDimPlot().

References

Skinnider, M.A., Squair, J.W., Kathe, C., et al. (2021). Cell type prioritization in single-cell data. Nature Biotechnology, 39, 30-34. doi:10.1038/s41587-020-0605-1

Examples

data(panc8_sub)
panc8_sub <- subset(panc8_sub, subset = tech %in% c("celseq", "celseq2"))
panc8_sub <- standard_scop(panc8_sub, verbose = FALSE)
panc8_sub <- RunAugur(
  panc8_sub,
  celltype.by = "celltype",
  label.by = "tech",
  n_subsamples = 5,
  subsample_size = 20,
  min_cells = 20,
  cores = 1,
  verbose = FALSE,
  rf_params = list(
    trees = 20,
    mtry = 2,
    min_n = NULL,
    importance = "accuracy"
  )
)

panc8_sub@tools$Augur$AUC
FeatureDimPlot(
  panc8_sub,
  features = "augur_auc",
  reduction = "StandardUMAP2D",
  bg_cutoff = -Inf
)