query.methodsMeanExpression: Runs multiple DE methods

In jhsiao999/ashbun: Methods and evaluation tools for single cell RNA-seq count data

Usage

query.methodsMeanExpression(counts, counts_normed, condition, libsize_factors,
  methodsMeanExpression = c("DESeq2", "limmaVoom", "edgeR", "BPSC", "MAST",
  "ROTS"))

Arguments

counts	Gene by sample expression count matrix (G by N). Use filtered count data.
counts_normed	Normalized expression count matrix (typicall CPM with normlized library size).
condition	Binary vector of length N indicating sample biological condition.
libsize_factors	Numeric vector of scale factors for library size factors.
methodsMeanExpression	Chararacter vector of evaluted methods. To run all methods, use c("DESeq2", "limmaVoom", "edgeR","BPSC", "MAST", "ROTS")
control	pseudocount Default .5. For limmaVoom and MAST. If NULL, then do not use pseudocount.

Value

pvalues data.frame of significance values. Columns corresond to input methods.

Author(s)

Chiaowen Joyce Hsiao

Examples

ipsc_eset <- get(load(system.file("testdata", "HumanTungiPSC.rda", package = "ashbun")))
counts <- exprs(ipsc_eset)[sample(nrow(exprs(ipsc_eset)), 500), ]
condition <- pData(ipsc_eset)$replicate

----- Step 1: filtering
counts_filtered <- filter.excludeAllZeros(counts)
featuresToInclude <- filterFeatures.fractionExpressed(counts_filtered,
                                                     thresholdDetection = 1,
                                                     fractionExpressed = .01)$index_filter

samplesToInclude <-  filterSamples.fractionExpressed(counts_filtered,
                                                     thresholdDetection = 1,
                                                     fractionExpressed = .01)$index_filter

counts_filtered <- counts_filtered[featuresToInclude, samplesToInclude]

---- Step 2: compute library size factors
libsize_factors <- normalize.scran(counts = counts_filtered)$libsize_factors
counts_normed <- normalize.cpm(counts_filtered, libsize_factors)$cpm

---- Step 3: run DE methods
pvals_list <- query.methodsMeanExpression(counts = counts_filtered,
                                          counts_normed = counts_normed,
                                          condition = condition_filtered,
                                          libsize_factors = libsize_factors,
                                          methodsMeanExpression = c("limmaVoom",
                                                                    "DESeq2",
                                                                    "edgeR",
                                                                    "MAST"))

jhsiao999/ashbun documentation built on May 8, 2019, 11:17 p.m.