Attempt to recover raw counts from the normalized matrix

Usage

RecoverCounts(
  srt,
  assay = NULL,
  trans = c("expm1", "exp", "none"),
  min_count = c(1, 2, 3),
  tolerance = 0.1,
  sf = NULL,
  verbose = TRUE
)

Arguments

srt: A Seurat object.
assay: Which assay to use. If NULL, the default assay of the Seurat object will be used. When the object also contains ChromatinAssay, the default assay and additional ChromatinAssay will be preprocessed sequentially.
trans: The transformation function to applied when data is presumed to be log-normalized.
min_count: Minimum UMI count of genes.
tolerance: When recovering the raw counts, the nCount of each cell is theoretically calculated as an integer. However, due to decimal point preservation during normalization, the calculated nCount is usually a floating point number close to the integer. The tolerance is its difference from the integer. Default is 0.1
sf: Set the scaling factor manually.
verbose: Whether to print the message. Default is TRUE.

Examples

data(pancreas_sub)
pancreas_sub <- standard_scop(pancreas_sub)
#> ℹ [2026-04-26 01:52:58] Start standard processing workflow...
#> ℹ [2026-04-26 01:52:58] Checking a list of <Seurat>...
#> ! [2026-04-26 01:52:59] Data 1/1 of the `srt_list` is "unknown"
#> ℹ [2026-04-26 01:52:59] Perform `NormalizeData()` with `normalization.method = 'LogNormalize'` on 1/1 of `srt_list`...
#> ℹ [2026-04-26 01:53:01] Perform `Seurat::FindVariableFeatures()` on 1/1 of `srt_list`...
#> ℹ [2026-04-26 01:53:01] Use the separate HVF from `srt_list`
#> ℹ [2026-04-26 01:53:01] Number of available HVF: 2000
#> ℹ [2026-04-26 01:53:02] Finished check
#> ℹ [2026-04-26 01:53:02] Perform `Seurat::ScaleData()`
#> ℹ [2026-04-26 01:53:02] Perform pca linear dimension reduction
#> ℹ [2026-04-26 01:53:03] Use stored estimated dimensions 1:20 for Standardpca
#> ℹ [2026-04-26 01:53:03] Perform `Seurat::FindClusters()` with `cluster_algorithm = 'louvain'` and `cluster_resolution = 0.6`
#> ℹ [2026-04-26 01:53:03] Reorder clusters...
#> ℹ [2026-04-26 01:53:03] Skip `log1p()` because `layer = data` is not "counts"
#> ℹ [2026-04-26 01:53:04] Perform umap nonlinear dimension reduction
#> ℹ [2026-04-26 01:53:04] Perform umap nonlinear dimension reduction using Standardpca (1:20)
#> ℹ [2026-04-26 01:53:08] Perform umap nonlinear dimension reduction using Standardpca (1:20)
#> ✔ [2026-04-26 01:53:13] Standard processing workflow completed
raw_counts <- GetAssayData5(
  pancreas_sub,
  assay = "RNA",
  layer = "counts"
)

# Normalized the data
pancreas_sub <- Seurat::NormalizeData(pancreas_sub)
#> Normalizing layer: counts

# Now replace counts with the log-normalized data matrix
data <- GetAssayData5(
  pancreas_sub,
  assay = "RNA",
  layer = "data"
)
new_pancreas_sub <- SeuratObject::SetAssayData(
  object = pancreas_sub,
  layer = "counts",
  new.data = data,
  assay = "RNA"
)
# Recover the counts and compare with the raw counts matrix
pancreas_sub <- RecoverCounts(new_pancreas_sub)
#> ℹ [2026-04-26 01:53:15] Data type is log-normalized
#> ℹ [2026-04-26 01:53:15] The data is presumed to be log-normalized
#> ℹ [2026-04-26 01:53:15] Perform "expm1" on the raw data
new_counts <- GetAssayData5(
  pancreas_sub,
  assay = "RNA",
  layer = "counts"
)
identical(raw_counts, new_counts)
#> [1] TRUE

Attempt to recover raw counts from the normalized matrix

Usage

Arguments

See also

Examples