tests/testthat/test_exposure_related_functions.R

In mSigTools: Mutational Signature Analysis Tools

test_that("Read and write exposure", {
  file <- "testdata/tiny.exposure.csv"
  x <- read_exposure(file)
  tfile <- tempfile()
  write_exposure(x, tfile)
  reread.x <- read_exposure(tfile)
  expect_equal(x, reread.x)

  x <- read_exposure(file, check.names = FALSE)
  write_exposure(x, tfile)
  x2 <- read_exposure(tfile, check.names = FALSE)
  expect_equal(x, x2)

  file2 <- "testdata/tiny.exposure.dup.csv"
  expect_error(x <- read_exposure(file2, check.names = FALSE))
})

test_that("plot_exposure_internal function", {
  exposure <- read_exposure("testdata/synthetic.exposure.csv")
  old.par <- par(mar = c(3.9, 3.9, 1, 1))
  out <-
    plot_exposure_internal(sort_exposure(exposure),
      xlim = c(0, ncol(exposure) * 2.5),
      cex.legend = 0.5, main = "test", cex.main = 0.8,
      cex.xaxis = 0.6
    )
  expect_equal(out$plot.success, TRUE)

  # Only plot the first 30 samples
  out1 <- plot_exposure_internal(sort_exposure(exposure[, 1:30]),
    plot.proportion = TRUE, cex.legend = 0.5,
    main = "test1", cex.main = 0.8
  )
  expect_equal(out1$plot.success, TRUE)

  # Only plot one sample
  out2 <- plot_exposure_internal(sort_exposure(exposure[, 1, drop = FALSE]),
    xlim = c(0, 5), legend.x = 1.5,
    cex.legend = 0.4
  )
  expect_equal(out2$plot.success, TRUE)

  # Only plot selected samples
  out3 <-
    plot_exposure_internal(sort_exposure(exposure[, 3:6]), cex.legend = 0.5)
  expect_equal(out3$plot.success, TRUE)

  # We can change the position of legend relative to the bar plot by changing
  # xlim and legend.x
  out4 <-
    plot_exposure_internal(sort_exposure(exposure[, 3:6]),
      xlim = c(0, 25),
      cex.legend = 0.5, legend.x = 5.5
    )
  expect_equal(out4$plot.success, TRUE)

  # Plot exposure proportions rather than counts
  out5 <-
    plot_exposure_internal(sort_exposure(exposure[, 3:6]),
      xlim = c(0, 25),
      plot.proportion = TRUE, col = c("red", "blue"),
      cex.legend = 0.5, legend.x = 5.5
    )
  expect_equal(out5$plot.success, TRUE)

  # Change y axis labels
  out6 <-
    plot_exposure_internal(sqrt(sort_exposure(exposure)),
      xlim = c(0, ncol(exposure) * 2.5),
      cex.legend = 0.5, main = "test", cex.main = 0.8,
      yaxis.labels = round(seq(
        0, max(colSums(exposure)),
        max(colSums(exposure)) / 4
      ))
    )
  expect_equal(out6$plot.success, TRUE)

  graphics.off()
  par(old.par)
})

test_that("plot_exposure function", {
  exposure <- read_exposure("testdata/synthetic.exposure.csv")

  old.par <-
    par(mfcol = c(2, 1), mar = c(2.2, 3.9, 2, 1), oma = c(2, 0, 0, 0))
  out <- plot_exposure(
    exposure = sort_exposure(exposure[, 1:43]),
    main = "test", cex.main = 0.8, cex.legend = 0.3,
    cex.xaxis = 0.6
  )
  expect_equal(out$plot.success, TRUE)

  # Only plot the first 30 samples
  out1 <- plot_exposure(
    exposure = sort_exposure(exposure[, 1:30]), # Test a trick edge case
    main = "test1", cex.main = 0.8, cex.legend = 0.45
  )
  expect_equal(out1$plot.success, TRUE)

  # Only plot one sample
  out2 <- plot_exposure(
    exposure = sort_exposure(exposure[, 1, drop = FALSE]),
    samples.per.line = 4, cex.legend = 0.45,
    legend.x = 1.5
  )
  expect_equal(out2$plot.success, TRUE)

  # Only plot selected samples
  out3 <- plot_exposure(
    exposure = sort_exposure(exposure[, 3:6]),
    samples.per.line = 4, cex.legend = 0.45,
    legend.x = 4.5
  )
  expect_equal(out3$plot.success, TRUE)

  # Plot exposure proportions rather than counts
  out4 <- plot_exposure(
    exposure = sort_exposure(exposure[, 1:43]),
    plot.proportion = TRUE, cex.legend = 0.2
  )
  expect_equal(out4$plot.success, TRUE)

  # Only plot selected samples with selected signatures
  out5 <- plot_exposure(
    exposure = sort_exposure(exposure[1:4, 3:6]),
    samples.per.line = 4, plot.proportion = TRUE,
    col = c("red", "blue"), cex.legend = 0.45,
    legend.x = 4.5
  )
  expect_equal(out5$plot.success, TRUE)

  out6 <- plot_exposure(
    exposure = sort_exposure(exposure[, 3:6]),
    samples.per.line = 4, plot.proportion = TRUE,
    col = c("red", "blue"), cex.legend = 0.45,
    legend.x = 4.5
  )
  expect_equal(out6$plot.success, TRUE)


  # Change y axis labels
  out7 <-
    plot_exposure_internal(sqrt(sort_exposure(exposure)),
      xlim = c(0, ncol(exposure) * 2.5),
      cex.legend = 0.5, main = "test", cex.main = 0.8,
      yaxis.labels = round(seq(
        0, max(colSums(exposure)),
        max(colSums(exposure)) / 4
      ))
    )
  expect_equal(out7$plot.success, TRUE)
  graphics.off()

  par(old.par)
})

test_that("plot_exposure_to_pdf function", {
  exposure <- read_exposure("testdata/synthetic.exposure.csv")

  out <- plot_exposure_to_pdf(exposure, file = file.path(tempdir(), "test.pdf"))
  expect_equal(out$plot.success, TRUE)

  # Plot the mutations in sqrt scale
  out0 <- plot_exposure_to_pdf(
    exposure = sqrt(sort_exposure(exposure)),
    file = file.path(tempdir(), "test0.pdf"),
    ylab = "Number of mutations (square root)",
    cex.xaxis = 0.9
  )
  expect_equal(out0$plot.success, TRUE)

  # Plot the mutations in sqrt scale and change y axis labels
  out0.1 <- plot_exposure_to_pdf(
    exposure = sqrt(sort_exposure(exposure)),
    file = file.path(tempdir(), "test0.1.pdf"),
    ylab = "Number of mutations (square root)",
    ylim = c(0, max(colSums(sqrt(exposure)))),
    yaxis.labels = round(seq(
      0, max(colSums(exposure)),
      max(colSums(exposure)) / 4
    ))
  )
  expect_equal(out0.1$plot.success, TRUE)

  # Set ylim for the plots
  out1 <- plot_exposure_to_pdf(
    exposure = sort_exposure(exposure),
    file = file.path(tempdir(), "test1.pdf"),
    ylim = c(0, max(colSums(exposure)))
  )
  expect_equal(out1$plot.success, TRUE)

  # Only plot the first 30 samples
  out2 <- plot_exposure_to_pdf(
    exposure = sort_exposure(exposure[, 1:30]),
    file = file.path(tempdir(), "test2.pdf")
  ) # Test a trick edge case
  expect_equal(out2$plot.success, TRUE)

  # Only plot selected samples, we can change mfrow, cex.legend and legend.x
  # to make the plot look nicer.
  out3 <- plot_exposure_to_pdf(
    exposure = sort_exposure(exposure[, 1:10]),
    file = file.path(tempdir(), "test3.pdf"),
    mfrow = c(2, 2), cex.legend = 1.2,
    legend.x = 5.5,
    samples.per.line = 5
  )
  expect_equal(out3$plot.success, TRUE)

  # Only plot selected samples with selected signatures
  out4 <- plot_exposure_to_pdf(
    exposure = exposure[1:5, 1:15],
    file = file.path(tempdir(), "test4.pdf"),
    mfrow = c(3, 3), cex.legend = 1.38,
    legend.x = 5.5, ylim = c(0, 3800),
    legend.y = 3300,
    samples.per.line = 5
  )
  expect_equal(out4$plot.success, TRUE)

  # Plot exposure proportions rather than counts
  out5 <- plot_exposure_to_pdf(
    exposure = sort_exposure(exposure),
    file = file.path(tempdir(), "test5.pdf"),
    plot.proportion = TRUE
  )
  expect_equal(out5$plot.success, TRUE)

  # Only plot selected samples with selected signatures
  out6 <- plot_exposure_to_pdf(
    exposure = exposure[1:5, 1:15],
    file = file.path(tempdir(), "test6.pdf"),
    mfrow = c(3, 3), cex.legend = 1.38,
    legend.x = 5.5,
    col = c("red", "blue", "orange", "green", "purple"),
    samples.per.line = 5, plot.proportion = TRUE
  )
  expect_equal(out6$plot.success, TRUE)
  dev.off()
  unlink(file.path(tempdir(), paste0("test", c("", 1:6), ".pdf")))
  unlink("Rplots.pdf")
})