Perform Scissor Screening Analysis

Identifies phenotype-associated cell subpopulations in single-cell data using regularized regression on matched bulk expression profiles.

Usage

DoScissor(
   path2load_scissor_cache = NULL,
   path2save_scissor_inputs = "Scissor_inputs.RData",
   matched_bulk,
   sc_data,
   phenotype,
   label_type = "scissor",
   alpha = c(0.05,NULL),
   cutoff = 0.2,
   scissor_family = c("gaussian", "binomial", "cox"),
   reliability_test = FALSE,
   reliability_test.n = 10,
   reliability_test.nfold = 10,
   cell_evaluation = FALSE,
   cell_evaluation.benchmark_data = "path_to_file.RData",
   cell_evaluation.FDR = 0.05,
   cell_evaluation.bootstrap_n = 100,
   verbose = TRUE,
   ...
)

Arguments

path2load_scissor_cache: Path to precomputed Scissor inputs (RData file). If provided, skips recomputation (default: NULL).
path2save_scissor_inputs: Path to save intermediate files (default: "Scissor_inputs.RData").
matched_bulk: Normalized bulk expression matrix (features × samples). Column names must match phenotype identifiers.
sc_data: Seurat object containing single-cell RNA-seq data.
phenotype: Clinical outcome data. Can be: - Vector: named with sample IDs - Data frame: with row names matching bulk columns
label_type: Character specifying phenotype label type (e.g., "SBS1", "time"), stored in scRNA_data@misc
alpha: Parameter used to balance the effect of the l1 norm and the network-based penalties. It can be a number or a searching vector. If alpha = NULL, a default searching vector is used. The range of alpha is between 0 and 1. A larger alpha lays more emphasis on the l1 norm.
cutoff: (default: 0.2). When alpha=NULL, the cutoff is used to determine the optimal alpha. Higher values increase specificity.
scissor_family: Model family for outcome type: - "gaussian": Continuous outcomes - "binomial": Binary outcomes (default) - "cox": Survival outcomes
reliability_test: Logical to perform stability assessment when scissor_alpha is specified as a value between 0 and 1.(default: TRUE).
reliability_test.n: Number of CV folds for reliability test (default: 10).
reliability_test.nfold: Cross-validation folds for reliability test (default: 10).
cell_evaluation: Logical to perform cell evaluation (default: FALSE).
cell_evaluation.benchmark_data: Path to benchmark data (RData file).
cell_evaluation.FDR: FDR threshold for cell evaluation (default: 0.05).
cell_evaluation.bootstrap_n: Number of bootstrap iterations for cell evaluation (default: 100).
verbose: Logical to print progress messages (default: TRUE).
...: Additional arguments passed to Scissor.v5.optimized.

Value

A list containing:

scRNA_data

A Seurat object with screened cells containing metadata:

scissor: "Positive"/"Negative"/"Neutral" classification
label_type: Outcome label used

scissor_result

Raw Scissor results

reliability_result

If reliability_test=TRUE, contains:

statistic: A value between 0 and 1
p: p-value of the test statistic
AUC_test_real: 10 values of AUC for real data
AUC_test_back: A list of AUC for background data

cell_evaluation

If cell_evaluation=TRUE, contains:

evaluation_res: A data.frame with some supporting information for each Scissor selected cell

LICENSE

Licensed under the GNU General Public License version 3 (GPL-3.0). A copy of the license is available at https://www.gnu.org/licenses/gpl-3.0.en.html.

References

Sun D, Guan X, Moran AE, Wu LY, Qian DZ, Schedin P, et al. Identifying phenotype-associated subpopulations by integrating bulk and single-cell sequencing data. Nat Biotechnol. 2022 Apr;40(4):527–38.

Examples

if (FALSE) { # \dontrun{
# Binary outcome example
res <- DoScissor(
  matched_bulk = bulk_matrix,
  sc_data = seurat_obj,
  phenotype = a_named_vector,
  scissor_family = "binomial"
)
} # }