tests/testthat/test-example_test.R

In jgilis/satuRn_jgilis: Scalable Analysis of Differential Transcript Usage for Bulk and Single-Cell RNA-sequencing Applications

context("StatModel construction")

test_that("StatModel construction", {
  ## Empty model
  expect_true(validObject(StatModel()))
  ## Default model type
  expect_identical(StatModel()@type, "fitError")
  ## Fully defined model
  mod <- StatModel(type = "glm",
                   params = list(x = 3, y = 7, b = 4),
                   varPosterior = c(0.1, 0.2, 0.3),
                   dfPosterior = c(6, 7, 8))    
  expect_true(validObject(mod))    
})

##############################################################

context("Statmodel accession") # 2/5 accessors tested

data("sumExp_example", package = "satuRn")

sumExp_example <- fitDTU(
  object = sumExp_example,
  formula = ~0+group,
  parallel = FALSE,
  BPPARAM = BiocParallel::bpparam(),
  verbose = TRUE
)

mod1 <- rowData(sumExp_example)[["fitDTUModels"]]$"ENSMUST00000165123"
mod_empty <- StatModel()

test_that("model parameters can be accessed",{
  
    # test getModel
    expect_identical(class(getModel(mod1)),"list")
    expect_equal(length(getModel(mod1)),4)
    expect_identical(class(getModel(mod_empty)),"list")
    expect_equal(length(getModel(mod_empty)),0)
    
    # test getCoef
    expect_identical(class(getCoef(mod1)),"numeric")
    expect_identical(length(getCoef(mod1)),length(levels(as.factor(colData(sumExp_example)$group))))
})

##############################################################

context("fitDTU for different input types")

# load data
data("sumExp_example", package = "satuRn")

test_that("different fitDTU assay types give same results", {
  
  # matrix input type
  sumExp_mat <- sumExp_example
  sumExp_mat <- fitDTU(
    object = sumExp_mat,
    formula = ~0+group,
    parallel = FALSE,
    BPPARAM = BiocParallel::bpparam(),
    verbose = TRUE
  )
  
  # DataFrame input type
  mat <- sumExp_example@assays@data$counts
  sumExp_df <- sumExp_example
  sumExp_df@assays@data$counts <- mat
  sumExp_df <- satuRn::fitDTU(
    object = sumExp_df,
    formula = ~0+group,
    parallel = FALSE,
    BPPARAM = BiocParallel::bpparam(),
    verbose = TRUE
  )
  
  # sparseMatrix input type
  sumExp_sparse <- sumExp_example
  sumExp_sparse@assays@data$counts <- Matrix(mat, sparse = TRUE)
  sumExp_sparse <- satuRn::fitDTU(
    object = sumExp_sparse,
    formula = ~0+group,
    parallel = FALSE,
    BPPARAM = BiocParallel::bpparam(),
    verbose = TRUE
  )
  
  # DelayedMatrix input type
  sumExp_delayed <- sumExp_example
  sumExp_delayed@assays@data$counts <- DelayedArray(mat)
  sumExp_delayed <- satuRn::fitDTU(
    object = sumExp_delayed,
    formula = ~0+group,
    parallel = FALSE,
    BPPARAM = BiocParallel::bpparam(),
    verbose = TRUE
  )
  
  expect_equal(rowData(sumExp_mat)[["fitDTUModels"]], rowData(sumExp_df)[["fitDTUModels"]])
  expect_equal(rowData(sumExp_mat)[["fitDTUModels"]], rowData(sumExp_sparse)[["fitDTUModels"]])
  expect_equal(rowData(sumExp_mat)[["fitDTUModels"]], rowData(sumExp_delayed)[["fitDTUModels"]])
})