tests/testthat/test-convert_molecular_formula_to_data_table.R

In ume: Ultrahigh-Resolution Mass Spectrometry Data Evaluation for Complex Organic Matter

# Test cases
test_that("convert_molecular_formula_to_data_table works correctly", {

  # Test valid molecular formulas
  molecular_formulas <- c("C10H23NO4", "C10H24N4O2S", "C6[13C2]H12[18O2]ONaCl")
  result <- convert_molecular_formula_to_data_table(molecular_formulas)

  # Test that the result is a data.table
  expect_true(is.data.table(result))

  # Test that we have the correct number of rows (3 formulas)
  expect_equal(nrow(result), 3)

  # Test that the 'mf' column exists and is of character type
  expect_true("mf" %in% colnames(result))
  expect_type(result$mf, "character")

  # Test that element counts are correctly calculated (you would need to verify the expected counts for these examples)
  expect_true(all(result$C >= 0))  # Checking that there are counts for Carbon (C)
  expect_true(all(result$H >= 0))  # Checking that there are counts for Hydrogen (H)

  # Test that the warning is raised for duplicates
  molecular_formulas_with_duplicates <- c("C10H23NO4", "C10H23NO4", "C6H12O6")
  expect_warning(
    convert_molecular_formula_to_data_table(molecular_formulas_with_duplicates),
    "duplicates identified"
  )

  # Test for invalid element (should stop)
  invalid_formula <- c("C10H23NO4Zz")  # Assuming 'Zz' is not a valid element
  expect_error(
    convert_molecular_formula_to_data_table(invalid_formula),
    "Some formulas contain invalid element/isotope symbols"
  )

  # Test isotopic notation handling
  isotopic_formula <- c("C6[13C2]H12[18O2]ONaCl")
  result_iso <- convert_molecular_formula_to_data_table(isotopic_formula)

  # Test that isotopic formulas are handled correctly
  expect_true("13C" %in% colnames(result_iso))
  expect_true("18O" %in% colnames(result_iso))

  # Test the 'mass' column to ensure it's calculated correctly
  expect_true("mass" %in% colnames(result))
  expect_true(all(result$mass > 0))  # Check that mass is positive

  # Test the output when 'table_format' is 'long'
  result_long <- convert_molecular_formula_to_data_table(molecular_formulas, table_format = "long")
  expect_true(is.data.table(result_long))

  # Test that 'long' format includes the 'element' and 'count' columns
  expect_true("symbol" %in% colnames(result_long))
  expect_true("count" %in% colnames(result_long))

})

# Test edge case: empty input vector
# test_that("convert_molecular_formula_to_data_table handles empty input", {
#   result_empty <- convert_molecular_formula_to_data_table(c())
#   expect_equal(nrow(result_empty), 0)
# })

# Test edge case: input with NA or empty strings
test_that("convert_molecular_formula_to_data_table handles NA or empty strings", {
  expect_error(convert_molecular_formula_to_data_table(c(NA, "")), "'mf' must be provided.")
})