test-normalise.R
In NACHO: NanoString Quality Control Dashboard

test_that("default settings", {
  res <- normalise(
    nacho_object = GSE74821
  )
  expect_s3_class(res, "nacho")
})

test_that("missing nacho", {
  expect_error(normalise())
})

test_that("missing field", {
  GSE74821$nacho <- NULL
  expect_error(normalise(GSE74821))
})

test_that("No POS_E", {
  GSE74821$nacho <- GSE74821$nacho[GSE74821$nacho$Name != "POS_E(0.5)", ]
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = NULL,
    housekeeping_predict = FALSE,
    housekeeping_norm = FALSE,
    normalisation_method = "GEO",
    n_comp = 10,
    remove_outliers = FALSE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("genes not null", {
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = c("RPLP0", "ACTB"),
    housekeeping_predict = FALSE,
    housekeeping_norm = FALSE,
    normalisation_method = "GEO",
    n_comp = 10,
    remove_outliers = FALSE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("predict TRUE", {
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = NULL,
    housekeeping_predict = TRUE,
    housekeeping_norm = FALSE,
    normalisation_method = "GEO",
    n_comp = 10,
    remove_outliers = FALSE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("norm TRUE", {
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = NULL,
    housekeeping_predict = FALSE,
    housekeeping_norm = TRUE,
    normalisation_method = "GEO",
    n_comp = 10,
    remove_outliers = FALSE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("method GLM", {
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = NULL,
    housekeeping_predict = FALSE,
    housekeeping_norm = FALSE,
    normalisation_method = "GLM",
    n_comp = 10,
    remove_outliers = FALSE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("n_comp 2", {
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = NULL,
    housekeeping_predict = FALSE,
    housekeeping_norm = FALSE,
    normalisation_method = "GEO",
    n_comp = 2,
    remove_outliers = FALSE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("n_comp 10", {
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = NULL,
    housekeeping_predict = FALSE,
    housekeeping_norm = FALSE,
    normalisation_method = "GEO",
    n_comp = 10,
    remove_outliers = FALSE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("outliers TRUE", {
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = NULL,
    housekeeping_predict = FALSE,
    housekeeping_norm = FALSE,
    normalisation_method = "GEO",
    n_comp = 10,
    remove_outliers = FALSE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})


test_that("Test outliers", {
  res <- normalise(
    nacho_object = GSE74821,
    remove_outliers = TRUE,
    outliers_thresholds = list(
      BD = c(0.15, 2.25),
      FoV = 95,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})


test_that("All LoD to zero", {
  GSE74821$nacho$LoD <- 0
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = c("RPLP0", "ACTB"),
    housekeeping_predict = FALSE,
    housekeeping_norm = FALSE,
    normalisation_method = "GEO",
    n_comp = 10,
    remove_outliers = FALSE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("All PC to zero", {
  GSE74821$nacho$PC <- 0
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = c("RPLP0", "ACTB"),
    housekeeping_predict = FALSE,
    housekeeping_norm = FALSE,
    normalisation_method = "GEO",
    n_comp = 10,
    remove_outliers = FALSE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("housekeeping_norm to FALSE and remove_outliers to TRUE", {
  GSE74821$nacho$PC <- 0
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = NULL,
    housekeeping_predict = FALSE,
    housekeeping_norm = FALSE,
    normalisation_method = "GEO",
    n_comp = 10,
    remove_outliers = TRUE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("housekeeping_norm to TRUE and remove_outliers to TRUE", {
  GSE74821$nacho$PC <- 0
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = NULL,
    housekeeping_predict = FALSE,
    housekeeping_norm = TRUE,
    normalisation_method = "GEO",
    n_comp = 10,
    remove_outliers = TRUE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("housekeeping_norm to TRUE and remove_outliers to TRUE", {
  GSE74821$nacho$PC <- 0
  res <- normalise(
    nacho_object = GSE74821,
    housekeeping_genes = NULL,
    housekeeping_predict = FALSE,
    housekeeping_norm = TRUE,
    normalisation_method = "GEO",
    n_comp = 10,
    remove_outliers = TRUE,
    outliers_thresholds = list(
      BD = c(0.1, 2.25),
      FoV = 75,
      LoD = 2,
      PCL = 0.95,
      Positive_factor = c(1 / 4, 4),
      House_factor = c(1 / 11, 11)
    )
  )
  expect_s3_class(res, "nacho")
})

test_that("housekeeping_norm to TRUE and remove_outliers to TRUE", {
  attr(GSE74821, "RCC_type") <- "something"
  expect_error(normalise(GSE74821))
})


test_that("Missing values in counts", {
  index <- sample(which(GSE74821$nacho$CodeClass == "Endogenous"), size = 25)
  GSE74821$nacho[index, "Count"] <- NA
  GSE74821$nacho[index, "Count_Norm"] <- NA
  expect_message(
    object = normalise(GSE74821, normalisation_method = "GEO"),
    regexp = "Missing values have been replaced with zeros for PCA"
  )
})

test_that("plexset", {
  expect_s3_class(
    object = {
      normalise(
        plexset_nacho,
        housekeeping_predict = TRUE,
        housekeeping_norm = TRUE
      )
    },
    class = "nacho"
  )
})

test_that("plexset GLM", {
  expect_s3_class(
    object = {
      normalise(
        plexset_nacho,
        housekeeping_predict = TRUE,
        housekeeping_norm = TRUE,
        normalisation_method = "GLM"
      )
    },
    class = "nacho"
  )
})

test_that("normalise() uses the n_comp it receives", {
  res <- suppressMessages(normalise(GSE74821, n_comp = 3))
  expect_identical(res[["n_comp"]], 3)
  expect_identical(nrow(res[["pc_sum"]]), 3L)
  expect_identical(
    grep("^PC[0-9]+$", names(res[["nacho"]]), value = TRUE),
    sprintf("PC%02d", 1:3)
  )
})

test_that("normalise() keeps the requested n_comp when it removes outliers", {
  thresholds <- GSE74821[["outliers_thresholds"]]
  thresholds[["FoV"]] <- 99.5
  res <- suppressMessages(normalise(
    GSE74821,
    n_comp = 4,
    remove_outliers = TRUE,
    outliers_thresholds = thresholds
  ))
  expect_identical(res[["n_comp"]], 4)
  expect_identical(nrow(res[["pc_sum"]]), 4L)
})

test_that("normalise() flags outliers against new thresholds", {
  thresholds <- GSE74821[["outliers_thresholds"]]
  thresholds[["BD"]] <- c(0.1, 0.2)
  res <- suppressMessages(normalise(GSE74821, outliers_thresholds = thresholds))
  expect_identical(
    res[["nacho"]][["is_outlier"]],
    check_outliers(res)[["nacho"]][["is_outlier"]]
  )
  expect_true(any(res[["nacho"]][["is_outlier"]]))
})

test_that("a panel without POS_E gives NA for PCL and LoD, not a failure", {
  no_pos_e <- GSE74821
  no_pos_e[["nacho"]] <- no_pos_e[["nacho"]][
    no_pos_e[["nacho"]][["Name"]] != "POS_E(0.5)",
  ]
  res <- suppressMessages(normalise(no_pos_e, normalisation_method = "GEO"))
  expect_true(all(is.na(res[["nacho"]][["PCL"]])))
  expect_true(all(is.na(res[["nacho"]][["LoD"]])))
  expect_false(anyNA(res[["nacho"]][["is_outlier"]]))
})