pbDS: pseudobulk DS analysis

In HelenaLC/ddSingleCell: Multi-sample multi-group scRNA-seq data analysis tools

pseudobulk DS analysis

Description

pbDS tests for DS after aggregating single-cell measurements to pseudobulk data, by applying bulk RNA-seq DE methods, such as edgeR, DESeq2 and limma.

Usage

pbDS(
  pb,
  method = c("edgeR", "DESeq2", "limma-trend", "limma-voom"),
  design = NULL,
  coef = NULL,
  contrast = NULL,
  min_cells = 10,
  filter = c("both", "genes", "samples", "none"),
  treat = FALSE,
  verbose = TRUE,
  BPPARAM = SerialParam(progressbar = verbose)
)

Arguments

pb	a SingleCellExperiment containing pseudobulks as returned by aggregateData.
method	a character string.
design	For methods "edegR" and "limma", a design matrix with row & column names(!) created with model.matrix; For "DESeq2", a formula with variables in colData(pb). Defaults to ~ group_id or the corresponding model.matrix.
coef	passed to glmQLFTest, contrasts.fit, results for method = "edgeR", "limma-x", "DESeq2", respectively. Can be a list for multiple, independent comparisons.
contrast	a matrix of contrasts to test for created with makeContrasts.
min_cells	a numeric. Specifies the minimum number of cells in a given cluster-sample required to consider the sample for differential testing.
filter	characterstring specifying whether to filter on genes, samples, both or neither.
treat	logical specifying whether empirical Bayes moderated-t p-values should be computed relative to a minimum fold-change threshold. Only applicable for methods "limma-x" (treat) and "edgeR" (glmTreat), and ignored otherwise.
verbose	logical. Should information on progress be reported?
BPPARAM	a BiocParallelParam object specifying how differential testing should be parallelized.

Value

a list containing

• a data.frame with differential testing results,

• a DGEList object of length nb.-clusters, and

• the design matrix, and contrast or coef used.

Author(s)

Helena L Crowell & Mark D Robinson

References

Crowell, HL, Soneson, C, Germain, P-L, Calini, D, Collin, L, Raposo, C, Malhotra, D & Robinson, MD: On the discovery of population-specific state transitions from multi-sample multi-condition single-cell RNA sequencing data. bioRxiv 713412 (2018). doi: https://doi.org/10.1101/713412

Examples

# simulate 5 clusters, 20% of DE genes
data(example_sce)
    
# compute pseudobulk sum-counts & run DS analysis
pb <- aggregateData(example_sce)
res <- pbDS(pb, method = "limma-trend")

names(res)
names(res$table)
head(res$table$stim$`B cells`)

# count nb. of DE genes by cluster
vapply(res$table$stim, function(u) 
  sum(u$p_adj.loc < 0.05), numeric(1))

# get top 5 hits for ea. cluster w/ abs(logFC) > 1
library(dplyr)
lapply(res$table$stim, function(u)
  filter(u, abs(logFC) > 1) %>% 
    arrange(p_adj.loc) %>% 
    slice(seq_len(5)))

HelenaLC/ddSingleCell documentation built on Feb. 21, 2023, 4:31 p.m.