tests/testthat/test_normalize_isobaric.R

In pmartR/pmartRqc: Panomics Marketplace - Quality Control and Statistical Analysis for Panomics Data

context('normalize: isobaric')

test_that('normalize_isobaric produces the correct output', {
  # Load the reduced peptide data frames ---------------------------------------

  load(system.file('testdata',
    'little_isodata.RData',
    package = 'pmartR'
  ))

  # Create a isobaricpepData object with the reduced data set.
  isodata <- as.isobaricpepData(
    e_data = edata,
    f_data = fdata,
    e_meta = emeta,
    edata_cname = "Peptide",
    fdata_cname = "Sample",
    emeta_cname = "Protein"
  )

  # Natural logate the data.
  isodata <- edata_transform(
    omicsData = isodata,
    data_scale = "log"
  )

  # Calculate the normalization standards --------------------------------------

  # Subtract the reference values from all other samples within each set.
  set1 <- isodata$e_data[, 2:4] - isodata$e_data[, 5]
  set2 <- isodata$e_data[, 6:8] - isodata$e_data[, 9]
  set3 <- isodata$e_data[, 10:12] - isodata$e_data[, 13]

  # Combine each set by column.
  norm_standard <- cbind(isodata$e_data[, 1], set1, set2, set3)
  names(norm_standard)[1] <- names(isodata$e_data)[1]

  # Test normalize_isobaric: isobaricnormRes -----------------------------------

  # Test error throwing
  err_1 <- "omicsData must be of the class 'isobaricpepData'"
  testthat::expect_error(normalize_isobaric(60), err_1)

  # Use the first specification for identifying the reference samples.
  spec1 <- normalize_isobaric(isodata,
    exp_cname = "Set",
    apply_norm = FALSE,
    channel_cname = "iTRAQ.Channel",
    refpool_channel = "116"
  )

  # Use the second specification for identifying the reference samples
  spec2 <- normalize_isobaric(isodata,
    exp_cname = "Set",
    apply_norm = FALSE,
    refpool_cname = "Reference",
    refpool_notation = "Yes"
  )

  # Inspect the class of the two specifications.
  expect_s3_class(spec1, "isobaricnormRes")
  expect_s3_class(spec2, "isobaricnormRes")

  # Make sure the two specifications produce the same output.
  expect_identical(spec1$e_data, spec2$e_data)
  # Change all columns in f_data to character vectors because depending on which
  # specification is used some columns will/won't be converted to a character
  # vector within the normalize_isobaric function.
  expect_identical(
    apply(spec1$f_data, 2, as.character),
    apply(spec2$f_data, 2, as.character)
  )

  # Check e_data and f_data are correct.
  expect_identical(
    spec1$e_data,
    isodata$e_data[, c(1, 5, 9, 13)]
  )
  expect_identical(
    apply(spec1$f_data, 2, as.character),
    apply(isodata$f_data[c(4, 8, 12), ], 2, as.character)
  )
  expect_identical(
    spec2$e_data,
    isodata$e_data[, c(1, 5, 9, 13)]
  )
  expect_identical(
    apply(spec2$f_data, 2, as.character),
    apply(isodata$f_data[c(4, 8, 12), ], 2, as.character)
  )

  # Scan the attributes of the isobaricnormRes objects for deficiencies.
  expect_identical(
    attr(spec1, "cnames"),
    list(
      edata_cname = "Peptide",
      fdata_cname = "Sample"
    )
  )
  expect_identical(
    attr(spec1, "isobaric_info"),
    list(
      exp_cname = "Set",
      channel_cname = "iTRAQ.Channel",
      refpool_channel = "116",
      refpool_cname = NULL,
      refpool_notation = NULL,
      norm_info = list(is_normalized = FALSE)
    )
  )
  expect_identical(
    attr(spec2, "cnames"),
    list(
      edata_cname = "Peptide",
      fdata_cname = "Sample"
    )
  )
  expect_identical(
    attr(spec2, "isobaric_info"),
    list(
      exp_cname = "Set",
      channel_cname = NULL,
      refpool_channel = NULL,
      refpool_cname = "Reference",
      refpool_notation = "Yes",
      norm_info = list(is_normalized = FALSE)
    )
  )

  # Test normalize_isobaric: apply normalization -------------------------------

  # Use the first specification for identifying the reference samples.
  spec1 <- normalize_isobaric(isodata,
    exp_cname = "Set",
    apply_norm = TRUE,
    channel_cname = "iTRAQ.Channel",
    refpool_channel = "116"
  )

  # Use the second specification for identifying the reference samples
  spec2 <- normalize_isobaric(isodata,
    exp_cname = "Set",
    apply_norm = TRUE,
    refpool_cname = "Reference",
    refpool_notation = "Yes"
  )

  # Ensure the normalization is the same between the two specifications.
  expect_identical(spec1$e_data, spec2$e_data)
  # Change all columns in f_data to character vectors because depending on which
  # specification is used some columns will/won't be converted to a character
  # vector within the normalize_isobaric function.
  expect_identical(
    apply(spec1$f_data, 2, as.character),
    apply(spec2$f_data, 2, as.character)
  )
  expect_identical(spec1$e_meta, spec2$e_meta)

  # Compare the normalized data to the standard.
  expect_identical(spec1$e_data, norm_standard)
  expect_identical(spec2$e_data, norm_standard)

  # Change the "Group" column in f_data to "group" because the pre_flight
  # function does not allow any column in f_data to be named "Group".
  names(fdata)[5] <- "group"

  # Make sure f_data has the correct rows removed. Metamorphose all columns into
  # character vectors. This way the comparisons are apples to apples instead of
  # something like apples to zombies.
  expect_identical(
    apply(spec1$f_data, 2, as.character),
    apply(fdata[-c(4, 8, 12), ], 2, as.character)
  )
  expect_identical(
    apply(spec2$f_data, 2, as.character),
    apply(fdata[-c(4, 8, 12), ], 2, as.character)
  )

  # Have a looksie at the class.
  expect_s3_class(spec1, c("isobaricpepData", "pepData"))
  expect_s3_class(spec2, c("isobaricpepData", "pepData"))

  # See if the attributes that should not have changed did not change.
  expect_identical(
    attr(isodata, "cnames"),
    attr(spec1, "cnames")
  )
  expect_identical(
    attr(isodata, "cnames"),
    attr(spec2, "cnames")
  )
  expect_identical(
    attr(isodata, "meta_info"),
    attr(spec1, "meta_info")
  )
  expect_identical(
    attr(isodata, "meta_info"),
    attr(spec2, "meta_info")
  )
  expect_identical(
    attr(isodata, "filters"),
    attr(spec1, "filters")
  )
  expect_identical(
    attr(isodata, "filters"),
    attr(spec2, "filters")
  )

  # Sleuth around the data_info attribute.
  expect_equal(
    attr(spec1, "data_info"),
    list(
      data_scale_orig = "abundance",
      data_scale = "log",
      norm_info = list(is_normalized = FALSE),
      num_edata = length(unique(spec1$e_data[, 1])),
      num_miss_obs = sum(is.na(spec1$e_data)),
      prop_missing = (sum(is.na(spec1$e_data)) /
        prod(dim(spec1$e_data[, -1]))),
      num_samps = ncol(spec1$e_data[, -1]),
      data_types = NULL,
      batch_info = list(is_bc = FALSE)
    )
  )
  expect_equal(
    attr(spec2, "data_info"),
    list(
      data_scale_orig = "abundance",
      data_scale = "log",
      norm_info = list(is_normalized = FALSE),
      num_edata = length(unique(spec2$e_data[, 1])),
      num_miss_obs = sum(is.na(spec2$e_data)),
      prop_missing = (sum(is.na(spec2$e_data)) /
        prod(dim(spec2$e_data[, -1]))),
      num_samps = ncol(spec2$e_data[, -1]),
      data_types = NULL,
      batch_info = list(is_bc = FALSE)
    )
  )

  # Confirm the isobaric_info attributes are correct.
  expect_identical(
    attr(spec1, "isobaric_info"),
    list(
      exp_cname = "Set",
      channel_cname = "iTRAQ.Channel",
      refpool_channel = "116",
      refpool_cname = NULL,
      refpool_notation = NULL,
      norm_info = list(is_normalized = TRUE)
    )
  )
  expect_identical(
    attr(spec2, "isobaric_info"),
    list(
      exp_cname = "Set",
      channel_cname = NULL,
      refpool_channel = NULL,
      refpool_cname = "Reference",
      refpool_notation = "Yes",
      norm_info = list(is_normalized = TRUE)
    )
  )
})