Skip to contents

Attempt to recover raw counts from the normalized matrix

Source: R/RecoverCounts.R
RecoverCounts.Rd

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

Assay to use. NULL uses the default assay.

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.

sf

Set the scaling factor manually.

verbose

Whether to print the message. Default is TRUE.

See also

CheckDataType

Examples

data(pancreas_sub)
pancreas_sub <- RunStandardWorkflow(pancreas_sub)
#> ℹ [2026-08-30 04:46:40] Start standard processing workflow...
#> ℹ [2026-08-30 04:46:40] Checking a list of <Seurat>...
#> ! [2026-08-30 04:46:41] Data 1/1 of the `srt_list` is "unknown"
#> Warning: Data 1/1 of the `srt_list` is "unknown"
#> ℹ [2026-08-30 04:46:41] Perform `NormalizeData()` with `normalization.method = 'LogNormalize'` on 1/1 of `srt_list`...
#> ℹ [2026-08-30 04:46:41] Perform `FindVariableFeatures()` on 1/1 of `srt_list`...
#> ℹ [2026-08-30 04:46:41] Use the separate HVF from `srt_list`
#> ℹ [2026-08-30 04:46:41] Number of available HVF: 2000
#> ℹ [2026-08-30 04:46:41] Finished check
#> ℹ [2026-08-30 04:46:41] Perform `ScaleData()`
#> ℹ [2026-08-30 04:46:41] Perform pca linear dimension reduction
#> ℹ [2026-08-30 04:46:41] Use stored estimated dimensions 1:23 for Standardpca
#> ℹ [2026-08-30 04:46:42] Perform `Seurat::FindClusters()` with `cluster_algorithm = 'louvain'` and `cluster_resolution = 0.6`
#> ℹ [2026-08-30 04:46:42] Reorder clusters...
#> ℹ [2026-08-30 04:46:42] Skip `log1p()` because `layer = data` is not "counts"
#> ℹ [2026-08-30 04:46:42] Perform umap nonlinear dimension reduction
#> ✔ [2026-08-30 04:46:49] 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 [2026-08-30 04:46:51] Data type is log-normalized
#> ℹ [2026-08-30 04:46:51] The data is presumed to be log-normalized
#> ℹ [2026-08-30 04:46:51] Perform "expm1" on the raw data
new_counts <- GetAssayData5(
  pancreas_sub,
  assay = "RNA",
  layer = "counts"
)
identical(raw_counts, new_counts)
#> [1] TRUE