Differential gene test

Perform differential expression testing on a Seurat object or a SummarizedExperiment bulk object. Users have the flexibility to specify custom cell groups, marker types, and various options for DE analysis.

Usage

RunDEtest(object = NULL, ...)

# S3 method for class 'Seurat'
RunDEtest(
  object,
  group.by = NULL,
  group1 = NULL,
  group2 = NULL,
  ident.1 = NULL,
  ident.2 = NULL,
  cells1 = NULL,
  cells2 = NULL,
  cells.1 = NULL,
  cells.2 = NULL,
  features = NULL,
  feature_type = c("gene", "peak", "cCRE"),
  markers_type = c("all", "paired", "conserved", "disturbed"),
  grouping.var = NULL,
  meta.method = c("maximump", "minimump", "wilkinsonp", "meanp", "sump", "votep"),
  test.use = "wilcox",
  only.pos = TRUE,
  fc.threshold = 1.5,
  logfc.threshold = NULL,
  base = 2,
  pseudocount.use = 1,
  mean.fxn = NULL,
  min.pct = 0.1,
  min.diff.pct = -Inf,
  max.cells.per.ident = Inf,
  latent.vars = NULL,
  min.cells.feature = 3,
  min.cells.group = 3,
  norm.method = "LogNormalize",
  sample_col = NULL,
  condition_col = NULL,
  min.cells.sample = 10,
  bulk_assay = "counts",
  p.adjust.method = "bonferroni",
  layer = "data",
  assay = NULL,
  seed = 11,
  verbose = TRUE,
  cores = 1,
  design = NULL,
  contrast = NULL,
  srt = NULL,
  ...
)

# S3 method for class 'SummarizedExperiment'
RunDEtest(
  object,
  group.by = NULL,
  group1 = NULL,
  group2 = NULL,
  cells1 = NULL,
  cells2 = NULL,
  features = NULL,
  feature_type = c("gene", "peak", "cCRE"),
  markers_type = c("all", "paired", "conserved", "disturbed"),
  grouping.var = NULL,
  meta.method = c("maximump", "minimump", "wilkinsonp", "meanp", "sump", "votep"),
  test.use = "edgeR",
  only.pos = TRUE,
  fc.threshold = 1.5,
  base = 2,
  pseudocount.use = 1,
  mean.fxn = NULL,
  min.pct = 0.1,
  min.diff.pct = -Inf,
  max.cells.per.ident = Inf,
  latent.vars = NULL,
  min.cells.feature = 3,
  min.cells.group = 3,
  norm.method = "LogNormalize",
  sample_col = NULL,
  condition_col = NULL,
  min.cells.sample = 10,
  bulk_assay = "counts",
  p.adjust.method = "bonferroni",
  layer = "counts",
  assay = NULL,
  seed = 11,
  verbose = TRUE,
  cores = 1,
  design = NULL,
  contrast = NULL,
  srt = NULL,
  ...
)

Arguments

object

A Seurat object or a SummarizedExperiment object.

...

Arguments passed to other functions.

group.by

A grouping variable in the dataset to define the groups or conditions for the differential test. If not provided, the function uses the "active.ident" variable in the Seurat object.

group1

A vector of cell IDs or a character vector specifying the cells that belong to the first group. If both group.by and group1 are provided, group1 takes precedence. For sample-level methods ("edgeR", "limma", "DESeq2", and "dream"), this parameter is interpreted as the first condition label.

group2

A vector of cell IDs or a character vector specifying the cells that belong to the second group. This parameter is only used when group.by or group1 is provided. For sample-level methods ("edgeR", "limma", "DESeq2", and "dream"), this parameter is interpreted as the second condition label.

ident.1, ident.2

Seurat-style aliases for group1 and group2.

cells1

A vector of cell IDs specifying the cells that belong to group1. If provided, group1 is ignored.

cells2

A vector of cell IDs specifying the cells that belong to group2. This parameter is only used when cells1 is provided.

cells.1, cells.2

Seurat-style aliases for cells1 and cells2.

features

A vector of feature names specifying the features to consider for the differential test. If not provided, all features in the dataset are considered.

feature_type

Feature type used for differential testing.

markers_type

A character value specifying the type of markers to find. Possible values are "all", "paired", "conserved", and "disturbed". Sample-level methods ("edgeR", "limma", "DESeq2", and "dream") currently support only "all".

grouping.var

A character value specifying the grouping variable for finding conserved or disturbed markers. This parameter is only used when markers_type is "conserved" or "disturbed".

meta.method

A character value specifying the method to use for combining p-values in the conserved markers test. Possible values are "maximump", "minimump", "wilkinsonp", "meanp", "sump", and "votep".

test.use

Differential testing method. "edgeR", "limma", "DESeq2", and "dream" run sample-level pseudobulk differential testing on Seurat input and bulk DE on SummarizedExperiment input.

only.pos

Only return positive markers (FALSE by default)

fc.threshold

A numeric value used to filter genes for testing based on their average fold change between/among the two groups.

logfc.threshold

Seurat-style log fold-change threshold. When provided, it is converted to fc.threshold = base^logfc.threshold.

base

The base with respect to which logarithms are computed.

pseudocount.use

Pseudocount to add to averaged expression values when calculating logFC. 1 by default.

mean.fxn

Function to use for fold change or average difference calculation. The default depends on the the value of fc.slot:

"counts" : difference in the log of the mean counts, with pseudocount.
"data" : difference in the log of the average exponentiated data, with pseudocount. This adjusts for differences in sequencing depth between cells, and assumes that "data" has been log-normalized.
"scale.data" : difference in the means of scale.data.

min.pct

only test genes that are detected in a minimum fraction of min.pct cells in either of the two populations. Meant to speed up the function by not testing genes that are very infrequently expressed. Default is 0.01

min.diff.pct

only test genes that show a minimum difference in the fraction of detection between the two groups. Set to -Inf by default

max.cells.per.ident

Down sample each identity class to a max number. Default is no downsampling. Not activated by default (set to Inf)

latent.vars

Variables to test, used only when test.use is one of 'LR', 'negbinom', 'poisson', or 'MAST'

min.cells.feature

Minimum number of cells expressing the feature in at least one of the two groups, currently only used for poisson and negative binomial tests

min.cells.group

Minimum number of cells in one of the groups

norm.method

Normalization method for fold change calculation when layer is 'data'. For a raw ChromatinAssay, RunDEtest() first applies Signac::RunTFIDF(); other assays default to "LogNormalize".

sample_col

Metadata column storing biological sample IDs. Required when test.use is a sample-level pseudobulk method on Seurat.

condition_col

Metadata column storing condition labels. Required when test.use is a sample-level pseudobulk method on Seurat, and required for SummarizedExperiment input.

min.cells.sample

Minimum number of cells required for a sample to be kept in sample-level pseudobulk testing. Applied independently within each group.by level (for example each cell type). Samples below the threshold are dropped before aggregation. Default 10. Set to 1 to keep every sample that has at least one cell. Ignored for cell-level tests and for SummarizedExperiment input.

bulk_assay

Assay name used as the bulk counts matrix for SummarizedExperiment input.

p.adjust.method

A character value specifying the method to use for adjusting p-values.

layer

Assay layer to use.

assay

Assay to use in differential expression testing

seed

Optional integer seed. When supplied, every input receives a deterministic independent L'Ecuyer-CMRG random-number stream, making results reproducible across worker counts and scheduling order. The caller's random number state is restored when the call finishes.

verbose

Whether to print the message. Default is TRUE.

cores

The number of worker processes to use for parallelization. Default is 1.

design

Optional one-sided formula for sample-level test.use = "edgeR", evaluated using cell metadata aggregated by sample_col, or bulk colData. Supply together with contrast, using exact argument names. Available for RNA/gene and ATAC peak/cCRE inputs; omitted arguments preserve legacy behavior. Every formula variable must be constant within an aggregate sample. For paired cell types, use distinct donor-by-cell-type sample IDs and a donor block such as ~ donor + cell_type. Missing combinations are never zero-filled. The model must have full rank and positive residual degrees of freedom, with at least two biological samples per selected condition. Formula offset() terms are not supported and are rejected explicitly.

contrast

Optional finite named numeric coefficient contrast for design. Names must match every fitted model-matrix column exactly, including zero-weight columns; order is aligned by name. Positive log-fold changes follow this vector, not necessarily group2 versus group1. Only one condition pair is fitted per group.by level. Invalid explicit models fail rather than being skipped.

srt

Deprecated alias for object; it will be removed in scop 1.0.0.

Details

Without design and contrast, existing defaults and comparison behavior are unchanged. Explicit edgeR models use filterByExpr(design = ...), TMM normalization, dispersion estimation, glmQLFit(robust = TRUE), and glmQLFTest(contrast = ...). Both p_val_adj (the requested adjustment) and p_val_BH use the full tested feature universe before fold-change display filtering. Set p.adjust.method = "BH" to also use BH for p_val_adj; its default remains "bonferroni". A fold-change display filter is not a minimum-effect-size hypothesis test.

Seurat results retain one receipt per group under object@tools[["DEtest_pseudobulk"]]$edgeR_models, or object@tools[[paste0("DEtest_", group.by)]]$edgeR_models. Each receipt contains full_results, input/tested/filtered feature IDs, raw pseudobulk counts, sample metadata and cell membership, cell counts, library sizes, normalization factors, design matrix, contrast, direction, residual degrees of freedom and dependency versions. Bulk receipts are under S4Vectors::metadata(object)$DEtest$methods$de_edgeR_qlf$details$groups, in each comparison's details$edgeR_model (without per-cell membership). Explicit peak/cCRE models require a ChromatinAssay or RangedSummarizedExperiment; full and displayed results include 1-based closed coordinates and supplied genome labels. No coordinates or count units are guessed from feature names. Cells and technical libraries must not be relabeled as independent biological replicates; metadata cannot establish independence.

Examples

data(pancreas_sub)
pancreas_sub <- RunStandardWorkflow(pancreas_sub)
#> ℹ [2026-11-12 05:30:09] Start standard processing workflow...
#> ℹ [2026-11-12 05:30:09] Checking a list of <Seurat>...
#> ! [2026-11-12 05:30:09] Data 1/1 of the `srt_list` is "unknown"
#> ℹ [2026-11-12 05:30:09] Perform `NormalizeData()` with `normalization.method = 'LogNormalize'` on 1/1 of `srt_list`...
#> ℹ [2026-11-12 05:30:09] Perform `FindVariableFeatures()` on 1/1 of `srt_list`...
#> ℹ [2026-11-12 05:30:09] Use the separate HVF from `srt_list`
#> ℹ [2026-11-12 05:30:09] Number of available HVF: 2000
#> ℹ [2026-11-12 05:30:09] Finished check
#> ℹ [2026-11-12 05:30:09] Perform `ScaleData()`
#> ℹ [2026-11-12 05:30:09] Perform pca linear dimension reduction
#> ℹ [2026-11-12 05:30:10] Use stored estimated dimensions 1:23 for Standardpca
#> ℹ [2026-11-12 05:30:10] Perform `Seurat::FindClusters()` with `cluster_algorithm = 'louvain'` and `cluster_resolution = 0.6`
#> ℹ [2026-11-12 05:30:10] Reorder clusters...
#> ℹ [2026-11-12 05:30:10] Skip `log1p()` because `layer = data` is not "counts"
#> ℹ [2026-11-12 05:30:10] Perform umap nonlinear dimension reduction
#> ✔ [2026-11-12 05:30:19] Standard processing workflow completed
pancreas_sub <- RunDEtest(
  object = pancreas_sub,
  group.by = "SubCellType",
  only.pos = FALSE
)
#> ℹ [2026-11-12 05:30:19] Data type is log-normalized
#> ℹ [2026-11-12 05:30:19] Start differential expression test
#> ℹ [2026-11-12 05:30:19] Find all markers(wilcox) among [1] 8 groups...
#> ✔ [2026-11-12 05:30:20] Differential expression test completed
AllMarkers <- dplyr::filter(
  pancreas_sub@tools$DEtest_SubCellType$AllMarkers_wilcox,
  p_val_adj < 0.05 & avg_log2FC > 1
)
ht1 <- GroupHeatmap(
  pancreas_sub,
  features = AllMarkers$gene,
  feature_split = AllMarkers$group1,
  group.by = "SubCellType"
)
#> ℹ [2026-11-12 05:30:21] The size of the heatmap is fixed because certain elements are not scalable.
#> ℹ [2026-11-12 05:30:21] The width and height of the heatmap are determined by the size of the current viewport.
#> ℹ [2026-11-12 05:30:21] If you want to have more control over the size, you can manually set the parameters 'width' and 'height'.
ht1$plot


TopMarkers <- AllMarkers |>
  dplyr::group_by(gene) |>
  dplyr::top_n(1, avg_log2FC) |>
  dplyr::group_by(group1) |>
  dplyr::top_n(3, avg_log2FC)
ht2 <- GroupHeatmap(
  pancreas_sub,
  features = TopMarkers$gene,
  feature_split = TopMarkers$group1,
  group.by = "SubCellType",
  show_row_names = TRUE
)
#> ℹ [2026-11-12 05:30:22] The size of the heatmap is fixed because certain elements are not scalable.
#> ℹ [2026-11-12 05:30:22] The width and height of the heatmap are determined by the size of the current viewport.
#> ℹ [2026-11-12 05:30:22] If you want to have more control over the size, you can manually set the parameters 'width' and 'height'.
ht2$plot


pancreas_sub <- RunDEtest(
  object = pancreas_sub,
  group.by = "SubCellType",
  markers_type = "paired",
  cores = 2
)
#> ℹ [2026-11-12 05:30:23] Data type is log-normalized
#> ℹ [2026-11-12 05:30:23] Start differential expression test
#> ℹ [2026-11-12 05:30:23] Find paired markers(wilcox) among [1] 8 groups...
#> ℹ [2026-11-12 05:30:23] Using 2 cores
#> ⠙ [2026-11-12 05:30:23] Running for 5 [1/28]              4% | ETA:  1m
#> ✔ [2026-11-12 05:30:23] Completed 28 tasks in 3.4s
#> 
#> ℹ [2026-11-12 05:30:23] Building results
#> ✔ [2026-11-12 05:30:27] Differential expression test completed
PairedMarkers <- dplyr::filter(
  pancreas_sub@tools$DEtest_SubCellType$PairedMarkers_wilcox,
  p_val_adj < 0.05 & avg_log2FC > 1
)
ht3 <- GroupHeatmap(
  pancreas_sub,
  features = PairedMarkers$gene,
  feature_split = PairedMarkers$group1,
  group.by = "SubCellType"
)
#> ℹ [2026-11-12 05:30:34] The size of the heatmap is fixed because certain elements are not scalable.
#> ℹ [2026-11-12 05:30:34] The width and height of the heatmap are determined by the size of the current viewport.
#> ℹ [2026-11-12 05:30:34] If you want to have more control over the size, you can manually set the parameters 'width' and 'height'.
ht3$plot


data(panc8_sub)
panc8_sub <- RunIntegration(
  panc8_sub,
  batch = "tech",
  integration_method = "Uncorrected"
)
#> ◌ [2026-11-12 05:30:36] Run integration workflow...
#> ℹ [2026-11-12 05:30:36] Split `srt_merge` into `srt_list` by "tech"
#> ℹ [2026-11-12 05:30:37] Checking a list of <Seurat>...
#> ! [2026-11-12 05:30:37] Data 1/5 of the `srt_list` is "unknown"
#> ℹ [2026-11-12 05:30:37] Perform `NormalizeData()` with `normalization.method = 'LogNormalize'` on 1/5 of `srt_list`...
#> ℹ [2026-11-12 05:30:37] Perform `FindVariableFeatures()` on 1/5 of `srt_list`...
#> ! [2026-11-12 05:30:37] Data 2/5 of the `srt_list` is "unknown"
#> ℹ [2026-11-12 05:30:37] Perform `NormalizeData()` with `normalization.method = 'LogNormalize'` on 2/5 of `srt_list`...
#> ℹ [2026-11-12 05:30:38] Perform `FindVariableFeatures()` on 2/5 of `srt_list`...
#> ! [2026-11-12 05:30:38] Data 3/5 of the `srt_list` is "unknown"
#> ℹ [2026-11-12 05:30:38] Perform `NormalizeData()` with `normalization.method = 'LogNormalize'` on 3/5 of `srt_list`...
#> ℹ [2026-11-12 05:30:38] Perform `FindVariableFeatures()` on 3/5 of `srt_list`...
#> ! [2026-11-12 05:30:38] Data 4/5 of the `srt_list` is "unknown"
#> ℹ [2026-11-12 05:30:38] Perform `NormalizeData()` with `normalization.method = 'LogNormalize'` on 4/5 of `srt_list`...
#> ℹ [2026-11-12 05:30:38] Perform `FindVariableFeatures()` on 4/5 of `srt_list`...
#> ! [2026-11-12 05:30:38] Data 5/5 of the `srt_list` is "unknown"
#> ℹ [2026-11-12 05:30:38] Perform `NormalizeData()` with `normalization.method = 'LogNormalize'` on 5/5 of `srt_list`...
#> ℹ [2026-11-12 05:30:38] Perform `FindVariableFeatures()` on 5/5 of `srt_list`...
#> ℹ [2026-11-12 05:30:38] Use the separate HVF from `srt_list`
#> ℹ [2026-11-12 05:30:38] Number of available HVF: 2000
#> ℹ [2026-11-12 05:30:38] Finished check
#> ℹ [2026-11-12 05:30:40] Perform Uncorrected integration
#> ℹ [2026-11-12 05:30:41] Perform `Seurat::ScaleData()`
#> ℹ [2026-11-12 05:30:42] Perform "pca" linear dimension reduction
#> ℹ [2026-11-12 05:30:43] Adjust neighbor k from 20 to 20 for small-sample clustering
#> ℹ [2026-11-12 05:30:43] Perform `Seurat::FindClusters()` with "louvain"
#> ℹ [2026-11-12 05:30:43] Reorder clusters...
#> ℹ [2026-11-12 05:30:44] Skip `log1p()` because `layer = data` is not "counts"
#> ℹ [2026-11-12 05:30:44] Perform umap nonlinear dimension reduction using Uncorrectedpca (1:20)
#> ℹ [2026-11-12 05:30:50] Perform umap nonlinear dimension reduction using Uncorrectedpca (1:20)
#> ℹ [2026-11-12 05:30:57] Perform umap nonlinear dimension reduction using Uncorrectedpca (1:20)
#> ✔ [2026-11-12 05:31:05] Uncorrected integration completed
CellDimPlot(
  panc8_sub,
  group.by = c("celltype", "tech")
)


panc8_sub <- RunDEtest(
  object = panc8_sub,
  group.by = "celltype",
  grouping.var = "tech",
  markers_type = "conserved",
  cores = 2
)
#> ℹ [2026-11-12 05:31:06] Data type is log-normalized
#> ℹ [2026-11-12 05:31:06] Start differential expression test
#> ℹ [2026-11-12 05:31:06] Find conserved markers(wilcox) among [1] 13 groups...
#> ℹ [2026-11-12 05:31:06] Using 2 cores
#> ⠙ [2026-11-12 05:31:06] Running for ductal [1/13]              8% | ETA:  2m
#> ⠹ [2026-11-12 05:31:06] Running for beta [4/13] ■■■         31% | ETA: 23s
#> ✔ [2026-11-12 05:31:06] Completed 13 tasks in 12.2s
#> 
#> ℹ [2026-11-12 05:31:06] Building results
#> ✔ [2026-11-12 05:31:18] Differential expression test completed
ConservedMarkers1 <- dplyr::filter(
  panc8_sub@tools$DEtest_celltype$ConservedMarkers_wilcox,
  p_val_adj < 0.05 & avg_log2FC > 1
)
ht4 <- GroupHeatmap(
  panc8_sub,
  layer = "data",
  features = ConservedMarkers1$gene,
  feature_split = ConservedMarkers1$group1,
  group.by = "tech",
  split.by = "celltype",
  within_groups = TRUE
)
#> `use_raster` is automatically set to TRUE for a matrix with more than
#> 2000 rows. You can control `use_raster` argument by explicitly setting
#> TRUE/FALSE to it.
#> 
#> Set `ht_opt$message = FALSE` to turn off this message.
#> ℹ [2026-11-12 05:31:23] The size of the heatmap is fixed because certain elements are not scalable.
#> ℹ [2026-11-12 05:31:23] The width and height of the heatmap are determined by the size of the current viewport.
#> ℹ [2026-11-12 05:31:23] If you want to have more control over the size, you can manually set the parameters 'width' and 'height'.
ht4$plot


panc8_sub <- RunDEtest(
  object = panc8_sub,
  group.by = "tech",
  grouping.var = "celltype",
  markers_type = "conserved",
  cores = 2
)
#> ℹ [2026-11-12 05:31:26] Data type is log-normalized
#> ℹ [2026-11-12 05:31:26] Start differential expression test
#> ℹ [2026-11-12 05:31:26] Find conserved markers(wilcox) among [1] 5 groups...
#> ℹ [2026-11-12 05:31:26] Using 2 cores
#> ⠙ [2026-11-12 05:31:26] Running for celseq [1/5] ■■          20% | ETA: 36s
#> ⠹ [2026-11-12 05:31:26] Running for smartseq2 [4/5] ■■■■■■■■    80% | ETA:  3s
#> ✔ [2026-11-12 05:31:26] Completed 5 tasks in 11.8s
#> 
#> ℹ [2026-11-12 05:31:26] Building results
#> ✔ [2026-11-12 05:31:38] Differential expression test completed
ConservedMarkers2 <- dplyr::filter(
  panc8_sub@tools$DEtest_tech$ConservedMarkers_wilcox,
  p_val_adj < 0.05 & avg_log2FC > 1
)
ht4 <- GroupHeatmap(
  object = panc8_sub,
  layer = "data",
  features = ConservedMarkers2$gene,
  feature_split = ConservedMarkers2$group1,
  group.by = "tech",
  split.by = "celltype"
)
#> ℹ [2026-11-12 05:31:39] The size of the heatmap is fixed because certain elements are not scalable.
#> ℹ [2026-11-12 05:31:39] The width and height of the heatmap are determined by the size of the current viewport.
#> ℹ [2026-11-12 05:31:39] If you want to have more control over the size, you can manually set the parameters 'width' and 'height'.
ht4$plot


panc8_sub <- RunDEtest(
  object = panc8_sub,
  group.by = "celltype",
  grouping.var = "tech",
  markers_type = "disturbed",
  cores = 2
)
#> ℹ [2026-11-12 05:31:41] Data type is log-normalized
#> ℹ [2026-11-12 05:31:41] Start differential expression test
#> ℹ [2026-11-12 05:31:41] Find disturbed markers(wilcox) among [1] 13 groups...
#> ℹ [2026-11-12 05:31:41] Using 2 cores
#> ⠙ [2026-11-12 05:31:41] Running for ductal [1/13]              8% | ETA:  3m
#> ⠹ [2026-11-12 05:31:41] Running for beta [4/13] ■■■         31% | ETA: 38s
#> ✔ [2026-11-12 05:31:41] Completed 13 tasks in 18.7s
#> 
#> ℹ [2026-11-12 05:31:41] Building results
#> ✔ [2026-11-12 05:32:00] Differential expression test completed
DisturbedMarkers <- dplyr::filter(
  panc8_sub@tools$DEtest_celltype$DisturbedMarkers_wilcox,
  p_val_adj < 0.05 & avg_log2FC > 1 & var1 == "smartseq2"
)
ht5 <- GroupHeatmap(
  object = panc8_sub,
  layer = "data",
  features = DisturbedMarkers$gene,
  feature_split = DisturbedMarkers$group1,
  group.by = "celltype",
  split.by = "tech"
)
#> `use_raster` is automatically set to TRUE for a matrix with more than
#> 2000 rows. You can control `use_raster` argument by explicitly setting
#> TRUE/FALSE to it.
#> 
#> Set `ht_opt$message = FALSE` to turn off this message.
#> ℹ [2026-11-12 05:32:03] The size of the heatmap is fixed because certain elements are not scalable.
#> ℹ [2026-11-12 05:32:03] The width and height of the heatmap are determined by the size of the current viewport.
#> ℹ [2026-11-12 05:32:03] If you want to have more control over the size, you can manually set the parameters 'width' and 'height'.
ht5$plot


gene_specific <- names(which(table(DisturbedMarkers$gene) == 1))
DisturbedMarkers_specific <- DisturbedMarkers[
  DisturbedMarkers$gene %in% gene_specific,
]
ht6 <- GroupHeatmap(
  object = panc8_sub,
  layer = "data",
  features = DisturbedMarkers_specific$gene,
  feature_split = DisturbedMarkers_specific$group1,
  group.by = "celltype",
  split.by = "tech"
)
#> ℹ [2026-11-12 05:32:07] The size of the heatmap is fixed because certain elements are not scalable.
#> ℹ [2026-11-12 05:32:07] The width and height of the heatmap are determined by the size of the current viewport.
#> ℹ [2026-11-12 05:32:07] If you want to have more control over the size, you can manually set the parameters 'width' and 'height'.
ht6$plot


ht7 <- GroupHeatmap(
  object = panc8_sub,
  layer = "data",
  aggregate_fun = function(x) mean(expm1(x)) + 1,
  features = DisturbedMarkers_specific$gene,
  feature_split = DisturbedMarkers_specific$group1,
  group.by = "celltype",
  grouping.var = "tech",
  numerator = "smartseq2"
)
#> ! [2026-11-12 05:32:09] When 'grouping.var' is specified, 'exp_method' can only be 'log2fc'
#> ℹ [2026-11-12 05:32:10] The size of the heatmap is fixed because certain elements are not scalable.
#> ℹ [2026-11-12 05:32:10] The width and height of the heatmap are determined by the size of the current viewport.
#> ℹ [2026-11-12 05:32:10] If you want to have more control over the size, you can manually set the parameters 'width' and 'height'.
ht7$plot


cell_index <- ave(
  seq_along(pancreas_sub$CellType),
  pancreas_sub$CellType,
  FUN = seq_along
)
pancreas_sub[["sample"]] <- paste0(
  "S",
  (cell_index - 1) %% 4 + 1
)
pancreas_sub[["condition"]] <- ifelse(
  pancreas_sub$sample %in% c("S1", "S2"),
  "ctrl",
  "case"
)
pancreas_sub <- RunDEtest(
  object = pancreas_sub,
  group.by = "CellType",
  sample_col = "sample",
  condition_col = "condition",
  test.use = "limma",
  fc.threshold = 1,
  layer = "counts",
  only.pos = FALSE
)
#> ℹ [2026-11-12 05:32:11] Start sample-level differential testing
#> calcNormFactors has been renamed to normLibSizes
#> calcNormFactors has been renamed to normLibSizes
#> calcNormFactors has been renamed to normLibSizes
#> calcNormFactors has been renamed to normLibSizes
#> calcNormFactors has been renamed to normLibSizes
#> ✔ [2026-11-12 05:32:13] Sample-level differential testing completed
DEtestPlot(
  pancreas_sub,
  group.by = "CellType",
  test.use = "limma",
  group_use = "Ductal",
  plot_type = "volcano",
  x_metric = "avg_log2FC",
  y_metric = "p_val"
)


pancreas_sub <- RunDEtest(
  object = pancreas_sub,
  group.by = "CellType",
  sample_col = "sample",
  condition_col = "condition",
  test.use = "edgeR",
  layer = "counts",
  fc.threshold = 1,
  min.cells.sample = 10,
  only.pos = FALSE
)
#> ℹ [2026-11-12 05:32:13] Start sample-level differential testing
#> calcNormFactors has been renamed to normLibSizes
#> calcNormFactors has been renamed to normLibSizes
#> calcNormFactors has been renamed to normLibSizes
#> calcNormFactors has been renamed to normLibSizes
#> calcNormFactors has been renamed to normLibSizes
#> ✔ [2026-11-12 05:32:16] Sample-level differential testing completed
DEtestPlot(
  pancreas_sub,
  group.by = "CellType",
  test.use = "edgeR",
  group_use = "Ductal",
  plot_type = "volcano",
  x_metric = "avg_log2FC",
  y_metric = "p_val",
  DE_threshold = "abs(avg_log2FC) > log2(1.5) & p_val < 0.05"
)


data(islet_bulk)
bulk_out <- RunDEtest(
  object = islet_bulk,
  condition_col = "condition",
  group1 = "control",
  group2 = "bfa",
  test.use = "edgeR",
  only.pos = FALSE,
  fc.threshold = 1
)
#> calcNormFactors has been renamed to normLibSizes
DEtestPlot(
  bulk_out,
  test.use = "edgeR",
  plot_type = "volcano",
  x_metric = "avg_log2FC",
  y_metric = "p_val"
)


# Explicit paired sample-level model (synthetic biological-sample counts).
local({
  withr::local_preserve_seed()
  set.seed(11)
  samples <- data.frame(
    donor = factor(rep(paste0("D", 1:4), each = 2)),
    condition = factor(rep(c("control", "treated"), 4)),
    row.names = paste0("sample", 1:8)
  )
  counts <- matrix(rnbinom(200 * 8, mu = 60, size = 12),
    nrow = 200,
    dimnames = list(paste0("gene", 1:200), rownames(samples))
  )
  bulk <- SummarizedExperiment::SummarizedExperiment(
    assays = list(counts = counts), colData = samples
  )
  columns <- colnames(model.matrix(~ donor + condition, samples))
  contrast <- setNames(rep(0, length(columns)), columns)
  contrast["conditiontreated"] <- 1
  fitted <- RunDEtest(bulk,
    condition_col = "condition",
    group1 = "control", group2 = "treated", test.use = "edgeR",
    design = ~ donor + condition, contrast = contrast,
    only.pos = FALSE, fc.threshold = 1, p.adjust.method = "BH", verbose = FALSE
  )
  results <- S4Vectors::metadata(fitted)$DEtest$results
  stopifnot(all(results$p_val_BH >= 0 & results$p_val_BH <= 1))
})
#> calcNormFactors has been renamed to normLibSizes
# End explicit paired sample-level model.

Usage

Arguments

Details

See also

Examples