tests/testthat/test-ko2kegg_abundance.R

In ggpicrust2: Make 'PICRUSt2' Output Analysis and Visualization Easier

# Test ko2kegg_abundance function

test_that("ko2kegg_abundance works with valid data frame input", {
  mock_ko_data <- data.frame(
    function. = c("K00001", "K00002", "K00003"),
    Sample1 = c(10, 20, 30),
    Sample2 = c(15, 25, 35),
    stringsAsFactors = FALSE
  )

  result <- suppressMessages(ko2kegg_abundance(data = mock_ko_data))

  expect_s3_class(result, "data.frame")
  if (nrow(result) > 0) {
    expect_true(all(result >= 0))
    expect_equal(ncol(result), 2)
  }
})

test_that("ko2kegg_abundance handles file input correctly", {
  temp_file <- tempfile(fileext = ".tsv")
  mock_ko_data <- data.frame(
    function. = c("K00001", "K00002", "K00003"),
    Sample1 = c(10, 20, 30),
    Sample2 = c(15, 25, 35),
    stringsAsFactors = FALSE
  )
  write.table(mock_ko_data, temp_file, sep = "\t", row.names = FALSE)

  result <- suppressMessages(ko2kegg_abundance(file = temp_file))
  unlink(temp_file)

  expect_s3_class(result, "data.frame")
  if (nrow(result) > 0) {
    expect_true(all(result >= 0))
  }
})

test_that("ko2kegg_abundance uses data when both file and data are supplied", {
  bad_file <- tempfile(fileext = ".tsv")
  writeLines(c("bad\tS1", "not_ko\t1"), bad_file)
  on.exit(unlink(bad_file), add = TRUE)

  ko_to_kegg_reference <- ggpicrust2:::load_reference_data("ko_to_kegg")
  real_kos <- head(unique(ko_to_kegg_reference$ko_id), 3)
  valid_data <- data.frame(
    function. = real_kos,
    Sample1 = c(10, 20, 30),
    stringsAsFactors = FALSE
  )

  expect_warning(
    result <- ko2kegg_abundance(file = bad_file, data = valid_data),
    "Using data and ignoring file"
  )
  expect_s3_class(result, "data.frame")
})

test_that("ko2kegg_abundance suppresses progress output by default in non-interactive use", {
  ko_to_kegg_reference <- ggpicrust2:::load_reference_data("ko_to_kegg")
  real_kos <- head(unique(ko_to_kegg_reference$ko_id), 3)
  valid_data <- data.frame(
    function. = real_kos,
    Sample1 = c(10, 20, 30),
    stringsAsFactors = FALSE
  )

  out <- capture.output(invisible(ko2kegg_abundance(data = valid_data)), type = "output")
  expect_equal(out, character(0))
})

test_that("ko2kegg_abundance progress bar closes safely", {
  ko_to_kegg_reference <- ggpicrust2:::load_reference_data("ko_to_kegg")
  real_kos <- head(unique(ko_to_kegg_reference$ko_id), 3)
  valid_data <- data.frame(
    function. = real_kos,
    Sample1 = c(10, 20, 30),
    stringsAsFactors = FALSE
  )

  expect_no_error(
    suppressWarnings(capture.output(
      result <- ko2kegg_abundance(data = valid_data, progress = TRUE),
      type = "output"
    ))
  )
  expect_s3_class(result, "data.frame")
})

test_that("ko2kegg_abundance throws appropriate errors", {
  expect_error(ko2kegg_abundance(), "Please provide either a file or a data.frame")
  expect_error(ko2kegg_abundance(file = "test.pdf"), "File does not exist")
  expect_error(ko2kegg_abundance(data = list(a = 1)), "must be a data.frame")

  # Negative values
  negative_data <- data.frame(function. = c("K00001"), Sample1 = c(-1), stringsAsFactors = FALSE)
  expect_error(ko2kegg_abundance(data = negative_data), "Negative values")

  # Non-numeric columns
  non_numeric <- data.frame(function. = c("K00001"), Sample1 = c("text"), stringsAsFactors = FALSE)
  expect_error(ko2kegg_abundance(data = non_numeric), "non-numeric")
})

test_that("ko2kegg_abundance fails fast when input is not KO data (e.g. EC IDs)", {
  # EC-number-shaped input: passes the first-column name check (column named
  # "function") but no ID matches the KO format. Should stop at
  # validate_feature_ids instead of silently producing an empty matrix.
  ec_shaped_data <- data.frame(
    `function` = c("1.1.1.1", "1.1.1.2", "2.7.1.1"),
    Sample1 = c(10, 20, 30),
    Sample2 = c(15, 25, 35),
    check.names = FALSE,
    stringsAsFactors = FALSE
  )

  expect_error(
    suppressMessages(ko2kegg_abundance(data = ec_shaped_data)),
    "None of the feature IDs match expected KO format"
  )
})

test_that("ko2kegg_abundance fails fast when KO IDs are absent from the KEGG reference", {
  # KO IDs that pass format validation (K#####) but are not in the reference.
  # Use IDs outside the current KEGG space to trigger the all-zero branch.
  unknown_ko_data <- data.frame(
    function. = c("K99991", "K99992", "K99993"),
    Sample1 = c(10, 20, 30),
    Sample2 = c(15, 25, 35),
    stringsAsFactors = FALSE
  )

  expect_error(
    suppressMessages(ko2kegg_abundance(data = unknown_ko_data)),
    "No KO IDs in the input matched any KEGG pathway"
  )
})

test_that("ko2kegg_abundance preserves sample names and removes zero pathways", {
  ko_to_kegg_reference <- ggpicrust2:::load_reference_data("ko_to_kegg")
  real_kos <- head(unique(ko_to_kegg_reference$ko_id), 5)

  mock_ko_data <- data.frame(
    function. = real_kos,
    SampleA = c(100, 200, 150, 180, 120),
    SampleB = c(50, 100, 75, 90, 60),
    stringsAsFactors = FALSE
  )

  result <- suppressMessages(ko2kegg_abundance(data = mock_ko_data))

  if (nrow(result) > 0) {
    expect_equal(colnames(result), c("SampleA", "SampleB"))
    expect_true(all(rowSums(result) > 0))
  }
})

test_that("KEGG pathway filter removes non-pathway buckets", {
  ko_to_kegg_reference <- ggpicrust2:::load_reference_data("ko_to_kegg")
  filtered_reference <- ggpicrust2:::filter_kegg_reference_to_pathways(ko_to_kegg_reference)

  expect_false("ko01001" %in% filtered_reference$pathway_id)
  expect_false("ko99980" %in% filtered_reference$pathway_id)
  expect_false(any(grepl("^(09180|09190)\\b", filtered_reference$level1)))
})

test_that("prokaryote filter removes eukaryotic pathways and keeps microbial disease pathways", {
  ko_to_kegg_reference <- ggpicrust2:::load_reference_data("ko_to_kegg")
  pathway_reference <- ggpicrust2:::filter_kegg_reference_to_pathways(ko_to_kegg_reference)
  filtered_reference <- ggpicrust2:::filter_kegg_reference_for_prokaryotes(pathway_reference)

  expect_false("ko05200" %in% filtered_reference$pathway_id)
  expect_false("ko04910" %in% filtered_reference$pathway_id)
  expect_true("ko05130" %in% filtered_reference$pathway_id)
  expect_true("ko01501" %in% filtered_reference$pathway_id)

  expect_false(any(grepl("^09150\\b", filtered_reference$level1)))
  expect_false(any(
    grepl("^09160\\b", filtered_reference$level1) &
      !grepl("^(09171|09175)\\b", filtered_reference$level2)
  ))
  expect_true(any(grepl("^09171\\b", filtered_reference$level2)))
  expect_true(any(grepl("^09175\\b", filtered_reference$level2)))
})

test_that("ko2kegg_abundance applies pathway and prokaryote filters to output pathways", {
  ko_to_kegg_reference <- ggpicrust2:::load_reference_data("ko_to_kegg")
  real_kos <- head(unique(ko_to_kegg_reference$ko_id), 10)

  mock_ko_data <- data.frame(
    function. = real_kos,
    Sample1 = seq_along(real_kos),
    stringsAsFactors = FALSE
  )

  result <- suppressMessages(ko2kegg_abundance(data = mock_ko_data))

  expect_false("ko01001" %in% rownames(result))
  expect_false("ko99980" %in% rownames(result))
  expect_false("ko05200" %in% rownames(result))
})

test_that("filter_for_prokaryotes = FALSE keeps true eukaryotic pathways but not non-pathway buckets", {
  ko_to_kegg_reference <- ggpicrust2:::load_reference_data("ko_to_kegg")
  kos <- unique(c(
    ko_to_kegg_reference$ko_id[ko_to_kegg_reference$pathway_id == "ko05200"],
    ko_to_kegg_reference$ko_id[ko_to_kegg_reference$pathway_id == "ko99980"]
  ))

  mock_ko_data <- data.frame(
    function. = head(kos, 20),
    Sample1 = seq_len(min(20, length(kos))),
    stringsAsFactors = FALSE
  )

  result <- suppressMessages(
    ko2kegg_abundance(data = mock_ko_data, filter_for_prokaryotes = FALSE)
  )

  expect_true("ko05200" %in% rownames(result))
  expect_false("ko99980" %in% rownames(result))
})