tests/testthat/test-normalisation.R

In dynverse/scaterlegacy: Single-cell analysis toolkit for gene expression data in R

# Tests for normalisation methods

context("test expected usage")

test_that("normaliseExprs does not fail on input with zero-variance features", {
    data("sc_example_counts")
    data("sc_example_cell_info")
    pd <- new("AnnotatedDataFrame", data = sc_example_cell_info)
    example_sceset <- newSCESet(countData = sc_example_counts, phenoData = pd)
    expect_that(normaliseExprs(example_sceset, method = "none", 
                                feature_set = 1:100), is_a("SCESet"))
})

test_that("we can compute normalised expression values with TMM method", {
    data("sc_example_counts")
    data("sc_example_cell_info")
    pd <- new("AnnotatedDataFrame", data = sc_example_cell_info)
    example_sceset <- newSCESet(countData = sc_example_counts, phenoData = pd)
    keep_gene <- rowSums(counts(example_sceset)) > 0
    example_sceset <- example_sceset[keep_gene,]
    
    example_sceset <- normaliseExprs(example_sceset, method = "TMM", 
                                     feature_set = 1:100)
    
    expect_that(example_sceset, is_a("SCESet"))
})

test_that("we can compute normalised expression values with RLE method", {
    data("sc_example_counts")
    data("sc_example_cell_info")
    pd <- new("AnnotatedDataFrame", data = sc_example_cell_info)
    example_sceset <- newSCESet(countData = sc_example_counts, phenoData = pd)
    keep_gene <- rowSums(counts(example_sceset)) > 0
    example_sceset <- example_sceset[keep_gene,]
    
    example_sceset <- normaliseExprs(example_sceset, method = "RLE", 
                                     feature_set = 1:100)
    
    expect_that(example_sceset, is_a("SCESet"))
})

# test_that("we can compute normalised expression values with upperquartile 
#           method", {
#     data("sc_example_counts")
#     data("sc_example_cell_info")
#     pd <- new("AnnotatedDataFrame", data = sc_example_cell_info)
#     example_sceset <- newSCESet(countData = sc_example_counts + 1, 
#                                 phenoData = pd)
#     keep_gene <- rowSums(counts(example_sceset)) > 0
#     example_sceset <- example_sceset[keep_gene,]
#     example_sceset <- example_sceset[
#         matrixStats::rowVars(counts(example_sceset)) > 0,]
#     
#     example_sceset <- normaliseExprs(example_sceset, method = "upperquartile", 
#                                      feature_set = 1:200)
#     
#     expect_that(example_sceset, is_a("SCESet"))
# })

test_that("we can compute normalised expression values with none method", {
    data("sc_example_counts")
    data("sc_example_cell_info")
    pd <- new("AnnotatedDataFrame", data = sc_example_cell_info)
    example_sceset <- newSCESet(countData = sc_example_counts, phenoData = pd)
    keep_gene <- rowSums(counts(example_sceset)) > 0
    example_sceset <- example_sceset[keep_gene,]
    
    example_sceset <- normaliseExprs(example_sceset, method = "none", 
                                     feature_set = 1:100)
    
    expect_that(example_sceset, is_a("SCESet"))
})

test_that("we can compute normalised expression values with a design matrix", {
    data("sc_example_counts")
    data("sc_example_cell_info")
    pd <- new("AnnotatedDataFrame", data = sc_example_cell_info)
    example_sceset <- newSCESet(countData = sc_example_counts, phenoData = pd)
    keep_gene <- rowSums(counts(example_sceset)) > 0
    example_sceset <- calculateQCMetrics(example_sceset[keep_gene,], 
                                         feature_controls = 1:40)
    design <- model.matrix(~example_sceset$Cell_Cycle +
                               example_sceset$pct_counts_top_200_features +
                               example_sceset$pct_dropout +
                               example_sceset$total_features)
    example_sceset <- normaliseExprs(example_sceset, method = "none", 
                                     design = design)
    expect_that(example_sceset, is_a("SCESet"))
    
    example_sceset <- normaliseExprs(example_sceset, method = "TMM", 
                                     design = design)
    expect_that(example_sceset, is_a("SCESet"))
    
    example_sceset <- normaliseExprs(example_sceset, method = "RLE", 
                                     design = design)
    expect_that(example_sceset, is_a("SCESet"))
})

test_that("we can compute normalise the object", {
    data("sc_example_counts")
    data("sc_example_cell_info")
    pd <- new("AnnotatedDataFrame", data = sc_example_cell_info)
    example_sceset <- newSCESet(countData = sc_example_counts, phenoData = pd)
    keep_gene <- rowSums(counts(example_sceset)) > 0
    example_sceset <- example_sceset[keep_gene,]
    
    example_sceset <- normaliseExprs(example_sceset, method = "none", 
                                     feature_set = 1:100)
    ## normalize
    example_sceset <- normalize(example_sceset)
    expect_that(example_sceset, is_a("SCESet"))
    
    ## normalise
    example_sceset <- normalise(example_sceset)    
    expect_that(example_sceset, is_a("SCESet"))
    
    ## check error if no size factors
    sizeFactors(example_sceset) <- NULL

    expect_warning(normalize(example_sceset), "size factors were not defined")
    
})