One-step Pipeline • CellTFusion

library(CellTFusion)
#> 
#>

For convenience, CellTFusion() is an all-in-one wrapper that automates the entire pipeline — computing features, running intermediate analyses, and returning cell group scores — in a single call.

The individual steps (feature computation, cell group construction, statistical analysis) are described in their respective articles:

Running the full pipeline

library(CellTFusion)

raw.counts <- CellTFusion::raw.counts.tuto
traitdata  <- CellTFusion::traitdata.tuto

res <- CellTFusion(
  raw.counts     = raw.counts,
  normalized     = TRUE,
  coldata        = traitdata,
  trait          = "Best.Confirmed.Overall.Response",
  trait.positive = "CR",
  deconv_methods = c("Quantiseq", "Epidish"),
  file_name      = "TestRun",
  corr           = 0.7,
  pval           = 0.05,
  high_corr_groups = 0.85,
  return         = FALSE
)

Key parameters

Parameter	Description
`raw.counts`	Raw or normalized count matrix (genes × samples)
`normalized`	Set `TRUE` if counts are already log-normalized
`coldata`	Data frame of clinical metadata (optional)
`trait`	Column in `coldata` to use for supervised analysis
`trait.positive`	Positive class label for the trait
`deconv_methods`	Deconvolution algorithms to use
`corr`	Correlation threshold for grouping cell types
`pval`	P-value threshold for significance filters
`high_corr_groups`	Threshold for merging highly correlated cell groups
`return`	If `TRUE`, returns the result object instead of writing to disk

Output

CellTFusion() returns a named list containing:

Element	Description
`$Cell_groups`	Cell group scores and composition
`$Processed_deconvolution`	Reduced deconvolution feature matrix
`$TF_modules`	WTCNA network output
`$Pathways`	PROGENy pathway activity scores