tests/testthat/test-Simulations-methods.R

In crmPack: Object-Oriented Implementation of Dose Escalation Designs

skip_on_cran()

testthat::local_mocked_bindings(
  .DefaultSimulations = function(...) {
    readRDS(testthat::test_path("fixtures", "default_simulations.Rds"))
  }
)

testthat::local_mocked_bindings(
  .DefaultDualSimulations = function(...) {
    readRDS(testthat::test_path("fixtures", "default_dual_simulations.Rds"))
  }
)

testthat::local_mocked_bindings(
  .DefaultPseudoSimulations = function(...) {
    readRDS(testthat::test_path("fixtures", "default_pseudo_simulations.Rds"))
  }
)

testthat::local_mocked_bindings(
  .DefaultPseudoDualSimulations = function(...) {
    readRDS(testthat::test_path(
      "fixtures",
      "default_pseudo_dual_simulations.Rds"
    ))
  }
)

testthat::local_mocked_bindings(
  .DefaultPseudoDualFlexiSimulations = function(...) {
    readRDS(testthat::test_path(
      "fixtures",
      "default_pseudo_dual_flexi_simulations.Rds"
    ))
  }
)

# plot ----

## plot-GeneralSimulations ----

test_that("plot-GeneralSimulations works correctly", {
  mySims <- .DefaultSimulations()

  # Test default plot (includes both trajectory and dosesTried, so returns gtable)
  result <- plot(mySims)
  expect_s3_class(result, "gtable")
  expect_equal(dim(result), c(2, 1)) # Should have 2 rows, 1 column
  expect_true(all(c("heights", "widths", "grobs") %in% names(result)))

  # Test trajectory plot only
  result_trajectory <- plot(mySims, type = "trajectory")
  expect_s3_class(result_trajectory, "ggplot")
  expect_equal(result_trajectory$labels$x, "Patient")
  expect_equal(result_trajectory$labels$y, "Dose Level")
  expect_doppel("plot_generalSims_trajectory", result_trajectory)

  # Test dosesTried plot only
  result_doses <- plot(mySims, type = "dosesTried")
  expect_s3_class(result_doses, "ggplot")
  expect_equal(result_doses$labels$x, "Dose level")
  expect_equal(result_doses$labels$y, "Average proportion [%]")
  expect_doppel("plot_generalSims_dosesTried", result_doses)

  # Test both plot types explicitly
  result_both <- plot(mySims, type = c("trajectory", "dosesTried"))
  expect_s3_class(result_both, "gtable")
  expect_equal(dim(result_both), c(2, 1)) # Should have 2 rows, 1 column
})

test_that("plot-GeneralSimulations fails gracefully with bad input", {
  mySims <- .DefaultSimulations()

  expect_error(plot(mySims, type = "invalid_type"), "should be one of")
  # The function doesn't error with mixed valid/invalid types, it just filters out invalid ones
  result_mixed <- plot(mySims, type = c("trajectory", "invalid"))
  expect_s3_class(result_mixed, "ggplot")
  expect_error(plot(mySims, type = character(0)), "must be of length")
})

## plot-DualSimulations ----

test_that("plot-DualSimulations works correctly", {
  mySims <- .DefaultDualSimulations()

  # Test plots specific to DualSimulations
  result <- plot(mySims, type = c("trajectory", "sigma2W"))
  expect_s3_class(result, "gtable")
  expect_equal(dim(result), c(2, 1)) # Should have 2 rows, 1 column

  result_rho <- plot(mySims, type = "rho")
  expect_s3_class(result_rho, "ggplot")
  expect_doppel("plot_dualSims_rho", result_rho)
  expect_equal(result_rho$labels$x, "")
  expect_equal(result_rho$labels$y, "Correlation estimates")

  # Test sigma2W plot specifically
  result_sigma2w <- plot(mySims, type = "sigma2W")
  expect_s3_class(result_sigma2w, "ggplot")
  expect_doppel("plot_dualSims_sigma2w", result_sigma2w)
  expect_equal(result_sigma2w$labels$y, "Biomarker variance estimates")
})

# summary ----

## summary-GeneralSimulations ----

test_that("summary-GeneralSimulations works correctly", {
  mySims <- .DefaultSimulations()
  myTruth <- function(x) plogis(x)

  # Test summary with truth function
  result <- summary(mySims, truth = myTruth)
  expect_s4_class(result, "GeneralSimulationsSummary")

  # Check specific slot values for consistency
  expect_equal(result@target, c(0.2, 0.35))
  expect_equal(result@nsim, length(mySims@data))
  expect_true(is.numeric(result@dose_selected))
  expect_true(is.numeric(result@n_obs))
  expect_true(all(result@prop_dlts >= 0 & result@prop_dlts <= 1))
  expect_true(all(result@mean_tox_risk >= 0 & result@mean_tox_risk <= 1))
  expect_true(result@prop_at_target >= 0 & result@prop_at_target <= 1)
  expect_equal(length(result@dose_selected), result@nsim)

  # Test with custom target interval
  result_custom <- summary(mySims, truth = myTruth, target = c(0.15, 0.25))
  expect_s4_class(result_custom, "GeneralSimulationsSummary")
  expect_equal(result_custom@target, c(0.15, 0.25))
  expect_equal(result_custom@nsim, result@nsim) # Same number of simulations

  # Results may differ due to different target interval (but not guaranteed with all truth functions)
  expect_true(is.numeric(result_custom@prop_at_target))
})

test_that("summary-GeneralSimulations fails gracefully with bad input", {
  mySims <- .DefaultSimulations()
  myTruth <- function(x) plogis(x)

  # Test that missing truth function fails.
  expect_error(summary(mySims), "\"truth\" is missing")

  # Test ranges of target interval errors.
  expect_error(
    summary(mySims, truth = myTruth, target = c(0.5, 0.3)),
    "target[1] < target[2]",
    fixed = TRUE
  )
  expect_error(
    summary(mySims, truth = myTruth, target = c(-0.1, 0.3)),
    "not >= 0"
  )
  expect_error(
    summary(mySims, truth = myTruth, target = c(0.2, 1.1)),
    "not <= 1"
  )
})

# Add regression tests for specific expected values
test_that("summary-GeneralSimulations produces expected values", {
  mySims <- .DefaultSimulations()
  myTruth <- function(x) plogis(x - 3) # Simple truth function

  result <- summary(mySims, truth = myTruth, target = c(0.2, 0.35))

  # These values should remain stable across refactoring
  expect_equal(result@nsim, length(mySims@data))
  expect_true(result@dose_most_selected > 0) # Should select some dose
  expect_true(result@prop_at_target >= 0 && result@prop_at_target <= 1)

  # Check consistency of slot lengths
  expect_equal(length(result@dose_selected), result@nsim)
  expect_equal(length(result@tox_at_doses_selected), result@nsim)
  expect_equal(length(result@n_above_target), result@nsim)
})

## summary-Simulations ----

test_that("summary-Simulations works correctly", {
  mySims <- .DefaultSimulations()
  myTruth <- plogis

  result <- summary(mySims, truth = myTruth)
  expect_s4_class(result, "SimulationsSummary")
  expect_snapshot(result)

  # Check that it has additional slots compared to GeneralSimulationsSummary
  expect_true("fit_at_dose_most_selected" %in% slotNames(result))
  expect_true("mean_fit" %in% slotNames(result))
  expect_true("stop_report" %in% slotNames(result))
  expect_true("additional_stats" %in% slotNames(result))

  # Check specific slot values
  expect_true(is.numeric(result@fit_at_dose_most_selected))
  expect_equal(length(result@fit_at_dose_most_selected), result@nsim)
  expect_true(all(
    result@fit_at_dose_most_selected >= 0 &
      result@fit_at_dose_most_selected <= 1
  ))

  # Check mean_fit structure
  expect_true(is.list(result@mean_fit))
  expect_true("truth" %in% names(result@mean_fit))
  expect_true("average" %in% names(result@mean_fit))
  expect_true("lower" %in% names(result@mean_fit))
  expect_true("upper" %in% names(result@mean_fit))
  expect_equal(length(result@mean_fit$truth), length(result@dose_grid))
})

## summary-DualSimulations ----

test_that("summary-DualSimulations works correctly", {
  mySims <- .DefaultDualSimulations()
  myTruthTox <- function(dose) pnorm((dose - 60) / 10)
  myTruthBio <- function(dose) {
    pmax(0.1, pmin(0.95, 0.2 + 0.6 * (dose / 100)^0.5))
  }

  result <- summary(mySims, trueTox = myTruthTox, trueBiomarker = myTruthBio)
  expect_s4_class(result, "DualSimulationsSummary")
  expect_snapshot(result)

  # Check dual-specific slots
  expect_true("biomarker_fit_at_dose_most_selected" %in% slotNames(result))
  expect_true("mean_biomarker_fit" %in% slotNames(result))

  # Check biomarker fit values
  expect_true(is.numeric(result@biomarker_fit_at_dose_most_selected))
  expect_equal(length(result@biomarker_fit_at_dose_most_selected), result@nsim)

  # Check mean_biomarker_fit structure
  expect_true(is.list(result@mean_biomarker_fit))
  expect_true("truth" %in% names(result@mean_biomarker_fit))
  expect_true("average" %in% names(result@mean_biomarker_fit))
  expect_true("lower" %in% names(result@mean_biomarker_fit))
  expect_true("upper" %in% names(result@mean_biomarker_fit))
  expect_equal(
    length(result@mean_biomarker_fit$truth),
    length(result@dose_grid)
  )

  # Check that it inherits from SimulationsSummary
  expect_true("fit_at_dose_most_selected" %in% slotNames(result))
  expect_true("mean_fit" %in% slotNames(result))
})

# Report ----

test_that("Report class can be initialized", {
  # Create a simple test object
  test_obj <- new(
    "GeneralSimulationsSummary",
    target = c(0.2, 0.35),
    target_dose_interval = c(50, 100),
    nsim = 10L,
    prop_dlts = rep(0.25, 10),
    mean_tox_risk = rep(0.3, 10),
    dose_selected = rep(75, 10),
    dose_most_selected = 75,
    obs_tox_rate_at_dose_most_selected = 0.28,
    n_obs = rep(20L, 10),
    n_above_target = rep(2L, 10),
    tox_at_doses_selected = rep(0.32, 10),
    prop_at_target = 0.8,
    dose_grid = seq(25, 300, 25),
    placebo = FALSE
  )

  # Create Report object
  r <- Report$new(
    object = test_obj,
    df = as.data.frame(matrix(nrow = 1, ncol = 0)),
    dfNames = character()
  )

  expect_s4_class(r, "Report")
  expect_true(is(r$object, "GeneralSimulationsSummary"))
  expect_true(is.data.frame(r$df))
  expect_equal(nrow(r$df), 1)
  expect_equal(ncol(r$df), 0)
  expect_equal(length(r$dfNames), 0)
})

test_that("Report$dfSave adds columns correctly", {
  test_obj <- new(
    "GeneralSimulationsSummary",
    target = c(0.2, 0.35),
    target_dose_interval = c(50, 100),
    nsim = 10L,
    prop_dlts = rep(0.25, 10),
    mean_tox_risk = rep(0.3, 10),
    dose_selected = rep(75, 10),
    dose_most_selected = 75,
    obs_tox_rate_at_dose_most_selected = 0.28,
    n_obs = rep(20L, 10),
    n_above_target = rep(2L, 10),
    tox_at_doses_selected = rep(0.32, 10),
    prop_at_target = 0.8,
    dose_grid = seq(25, 300, 25),
    placebo = FALSE
  )

  r <- Report$new(
    object = test_obj,
    df = as.data.frame(matrix(nrow = 1, ncol = 0)),
    dfNames = character()
  )

  # Save first value
  result1 <- r$dfSave(10, "nsim")
  expect_equal(result1, 10)
  expect_equal(ncol(r$df), 1)
  expect_equal(r$dfNames, "nsim")

  # Save second value
  result2 <- r$dfSave(0.8, "prop_at_target")
  expect_equal(result2, 0.8)
  expect_equal(ncol(r$df), 2)
  expect_equal(r$dfNames, c("nsim", "prop_at_target"))
})

test_that("Report$report generates correct output with percentages", {
  test_obj <- new(
    "GeneralSimulationsSummary",
    target = c(0.2, 0.35),
    target_dose_interval = c(50, 100),
    nsim = 10L,
    prop_dlts = rep(0.25, 10),
    mean_tox_risk = rep(0.3, 10),
    dose_selected = rep(75, 10),
    dose_most_selected = 75,
    obs_tox_rate_at_dose_most_selected = 0.28,
    n_obs = rep(20L, 10),
    n_above_target = rep(2L, 10),
    tox_at_doses_selected = rep(0.32, 10),
    prop_at_target = 0.8,
    dose_grid = seq(25, 300, 25),
    placebo = FALSE
  )

  r <- Report$new(
    object = test_obj,
    df = as.data.frame(matrix(nrow = 1, ncol = 0)),
    dfNames = character()
  )

  # Capture output
  output <- capture.output(
    r$report("prop_dlts", "Proportion of DLTs", percent = TRUE, digits = 1)
  )

  expect_true(any(grepl("Proportion of DLTs", output)))
  expect_true(any(grepl("25", output))) # Should show 25% (0.25 * 100)
  expect_equal(ncol(r$df), 1)
  expect_equal(r$dfNames, "prop_dlts")
})

test_that("Report$report generates correct output without percentages", {
  test_obj <- new(
    "GeneralSimulationsSummary",
    target = c(0.2, 0.35),
    target_dose_interval = c(50, 100),
    nsim = 10L,
    prop_dlts = rep(0.25, 10),
    mean_tox_risk = rep(0.3, 10),
    dose_selected = rep(75, 10),
    dose_most_selected = 75,
    obs_tox_rate_at_dose_most_selected = 0.28,
    n_obs = rep(20L, 10),
    n_above_target = rep(2L, 10),
    tox_at_doses_selected = rep(0.32, 10),
    prop_at_target = 0.8,
    dose_grid = seq(25, 300, 25),
    placebo = FALSE
  )

  r <- Report$new(
    object = test_obj,
    df = as.data.frame(matrix(nrow = 1, ncol = 0)),
    dfNames = character()
  )

  # Capture output
  output <- capture.output(
    r$report("n_obs", "Number of observations", percent = FALSE, digits = 0)
  )

  expect_true(any(grepl("Number of observations", output)))
  expect_true(any(grepl("20", output))) # Should show 20 (not 2000%)
  expect_false(any(grepl("%", output))) # Should not contain %
})

test_that("Report$report respects custom quantiles", {
  test_obj <- new(
    "GeneralSimulationsSummary",
    target = c(0.2, 0.35),
    target_dose_interval = c(50, 100),
    nsim = 10L,
    prop_dlts = seq(0.1, 0.5, length.out = 10),
    mean_tox_risk = rep(0.3, 10),
    dose_selected = rep(75, 10),
    dose_most_selected = 75,
    obs_tox_rate_at_dose_most_selected = 0.28,
    n_obs = rep(20L, 10),
    n_above_target = rep(2L, 10),
    tox_at_doses_selected = rep(0.32, 10),
    prop_at_target = 0.8,
    dose_grid = seq(25, 300, 25),
    placebo = FALSE
  )

  r <- Report$new(
    object = test_obj,
    df = as.data.frame(matrix(nrow = 1, ncol = 0)),
    dfNames = character()
  )

  # Capture output with custom quantiles
  output <- capture.output(
    r$report(
      "prop_dlts",
      "Proportion of DLTs",
      percent = TRUE,
      digits = 1,
      quantiles = c(0.25, 0.75)
    )
  )

  expect_true(any(grepl("Proportion of DLTs", output)))
  # Should contain quantile values
  expect_true(any(grepl("\\(", output)))
  expect_true(any(grepl(",", output)))
})

# show ----

## show-GeneralSimulationsSummary ----

test_that("show-GeneralSimulationsSummary works correctly", {
  # Create simulation and summary.
  emptydata <- Data(doseGrid = c(1, 3, 5, 10, 15, 20, 25))
  model <- LogisticLogNormal(
    mean = c(-0.85, 1),
    cov = matrix(c(1, -0.5, -0.5, 1), nrow = 2),
    ref_dose = 56
  )

  design <- Design(
    model = model,
    nextBest = NextBestNCRM(
      target = c(0.2, 0.35),
      overdose = c(0.35, 1),
      max_overdose_prob = 0.25
    ),
    stopping = StoppingMinPatients(nPatients = 6),
    increments = IncrementsRelative(
      intervals = c(0, 20),
      increments = c(1, 0.33)
    ),
    cohort_size = CohortSizeConst(size = 3),
    data = emptydata,
    startingDose = 3
  )

  myTruth <- probFunction(model, alpha0 = 7, alpha1 = 8)
  options <- McmcOptions(
    burnin = 10,
    step = 2,
    samples = 20,
    rng_kind = "Mersenne-Twister",
    rng_seed = 123
  )

  mySims <- simulate(
    design,
    args = NULL,
    truth = myTruth,
    nsim = 2,
    seed = 819,
    mcmcOptions = options,
    parallel = FALSE
  )
  simSummary <- summary(mySims, truth = myTruth)

  # Test that show method works (produces output)
  expect_output(show(simSummary))

  # Show methods should print something
  result <- capture.output(show(simSummary))
  expect_true(length(result) > 0)
  expect_snapshot(show(simSummary))
})

## show-SimulationsSummary ----

test_that("show-SimulationsSummary works correctly", {
  emptydata <- Data(doseGrid = c(1, 3, 5, 10, 15, 20, 25))
  model <- LogisticLogNormal(
    mean = c(-0.85, 1),
    cov = matrix(c(1, -0.5, -0.5, 1), nrow = 2),
    ref_dose = 56
  )

  design <- Design(
    model = model,
    nextBest = NextBestNCRM(
      target = c(0.2, 0.35),
      overdose = c(0.35, 1),
      max_overdose_prob = 0.25
    ),
    stopping = StoppingMinPatients(nPatients = 6),
    increments = IncrementsRelative(
      intervals = c(0, 20),
      increments = c(1, 0.33)
    ),
    cohort_size = CohortSizeConst(size = 3),
    data = emptydata,
    startingDose = 3
  )

  myTruth <- probFunction(model, alpha0 = 7, alpha1 = 8)
  options <- McmcOptions(
    burnin = 10,
    step = 2,
    samples = 20,
    rng_kind = "Mersenne-Twister",
    rng_seed = 123
  )

  mySims <- simulate(
    design,
    args = NULL,
    truth = myTruth,
    nsim = 2,
    seed = 819,
    mcmcOptions = options,
    parallel = FALSE
  )
  simSummary <- summary(mySims, truth = myTruth)

  # Test that show method works (produces output)
  expect_output(show(simSummary))

  # Show methods should print something
  result <- capture.output(show(simSummary))
  expect_true(length(result) > 0)
  expect_snapshot(show(simSummary))
})

## show-DualSimulationsSummary ----

test_that("show-DualSimulationsSummary works correctly", {
  mySims <- .DefaultDualSimulations()
  myTruthTox <- function(dose) pnorm((dose - 60) / 10)
  myTruthBio <- function(dose) {
    pmax(0.1, pmin(0.95, 0.2 + 0.6 * (dose / 100)^0.5))
  }

  simSummary <- summary(
    mySims,
    trueTox = myTruthTox,
    trueBiomarker = myTruthBio
  )

  # Test that show method works (produces output)
  expect_output(show(simSummary))

  # Show methods should print something
  result <- capture.output(show(simSummary))
  expect_true(length(result) > 0)
  expect_snapshot(show(simSummary))

  # Check for specific content in the output
  expect_true(any(grepl("Summary of.*simulations", result)))
  expect_true(any(grepl("Target toxicity interval", result)))
  expect_true(any(grepl("biomarker", result, ignore.case = TRUE)))
  expect_true(any(grepl("Number of patients", result)))
  expect_true(any(grepl("Doses selected", result)))
})

# plot summary objects ----

## plot-GeneralSimulationsSummary ----

test_that("plot-GeneralSimulationsSummary works correctly", {
  emptydata <- Data(doseGrid = c(1, 3, 5, 10, 15, 20, 25))
  model <- LogisticLogNormal(
    mean = c(-0.85, 1),
    cov = matrix(c(1, -0.5, -0.5, 1), nrow = 2),
    ref_dose = 56
  )

  design <- Design(
    model = model,
    nextBest = NextBestNCRM(
      target = c(0.2, 0.35),
      overdose = c(0.35, 1),
      max_overdose_prob = 0.25
    ),
    stopping = StoppingMinPatients(nPatients = 6),
    increments = IncrementsRelative(
      intervals = c(0, 20),
      increments = c(1, 0.33)
    ),
    cohort_size = CohortSizeConst(size = 3),
    data = emptydata,
    startingDose = 3
  )

  myTruth <- probFunction(model, alpha0 = 7, alpha1 = 8)
  options <- McmcOptions(
    burnin = 10,
    step = 2,
    samples = 20,
    rng_kind = "Mersenne-Twister",
    rng_seed = 123
  )

  mySims <- simulate(
    design,
    args = NULL,
    truth = myTruth,
    nsim = 3,
    seed = 819,
    mcmcOptions = options,
    parallel = FALSE
  )
  simSummary <- summary(mySims, truth = myTruth)

  # Test different plot types
  result_n_obs <- plot(simSummary, type = "nObs")
  expect_s3_class(result_n_obs, "ggplot")
  expect_doppel("plot_generalSimsSummary_nObs", result_n_obs)

  result_dose_selected <- plot(simSummary, type = "doseSelected")
  expect_s3_class(result_dose_selected, "ggplot")
  expect_doppel("plot_generalSimsSummary_doseSelected", result_dose_selected)

  result_prop_dlts <- plot(simSummary, type = "propDLTs")
  expect_s3_class(result_prop_dlts, "ggplot")
  expect_doppel("plot_generalSimsSummary_propDLTs", result_prop_dlts)

  result_n_above_target <- plot(simSummary, type = "nAboveTarget")
  expect_s3_class(result_n_above_target, "ggplot")
  expect_doppel("plot_generalSimsSummary_nAboveTarget", result_n_above_target)

  # Test multiple plot types
  result_multiple <- plot(simSummary, type = c("nObs", "doseSelected"))
  expect_s3_class(result_multiple, "gtable")
})

## plot-SimulationsSummary ----

test_that("plot-SimulationsSummary works correctly", {
  emptydata <- Data(doseGrid = c(1, 3, 5, 10, 15, 20, 25))
  model <- LogisticLogNormal(
    mean = c(-0.85, 1),
    cov = matrix(c(1, -0.5, -0.5, 1), nrow = 2),
    ref_dose = 56
  )

  design <- Design(
    model = model,
    nextBest = NextBestNCRM(
      target = c(0.2, 0.35),
      overdose = c(0.35, 1),
      max_overdose_prob = 0.25
    ),
    stopping = StoppingMinPatients(nPatients = 6),
    increments = IncrementsRelative(
      intervals = c(0, 20),
      increments = c(1, 0.33)
    ),
    cohort_size = CohortSizeConst(size = 3),
    data = emptydata,
    startingDose = 3
  )

  myTruth <- probFunction(model, alpha0 = 7, alpha1 = 8)
  options <- McmcOptions(
    burnin = 10,
    step = 2,
    samples = 20,
    rng_kind = "Mersenne-Twister",
    rng_seed = 123
  )

  mySims <- simulate(
    design,
    args = NULL,
    truth = myTruth,
    nsim = 3,
    seed = 819,
    mcmcOptions = options,
    parallel = FALSE
  )
  simSummary <- summary(mySims, truth = myTruth)

  # Test meanFit plot (specific to SimulationsSummary)
  result_mean_fit <- plot(simSummary, type = "meanFit")
  expect_s3_class(result_mean_fit, "ggplot")
  expect_doppel("plot_simSimsSummary_meanFit", result_mean_fit)

  # Test combination with general plots
  result_multiple <- plot(simSummary, type = c("meanFit", "nObs"))
  expect_s3_class(result_multiple, "gtable")
})

## plot-DualSimulationsSummary ----

test_that("plot-DualSimulationsSummary works correctly", {
  mySims <- .DefaultDualSimulations()
  myTruthTox <- function(dose) pnorm((dose - 60) / 10)
  myTruthBio <- function(dose) {
    pmax(0.1, pmin(0.95, 0.2 + 0.6 * (dose / 100)^0.5))
  }

  simSummary <- summary(
    mySims,
    trueTox = myTruthTox,
    trueBiomarker = myTruthBio
  )

  # Test meanBiomarkerFit plot (specific to DualSimulationsSummary)
  result_mean_bio_fit <- plot(simSummary, type = "meanBiomarkerFit")
  expect_s3_class(result_mean_bio_fit, "ggplot")
  expect_doppel("plot_dualSimsSummary_meanBiomarkerFit", result_mean_bio_fit)
  expect_equal(result_mean_bio_fit$labels$x, "Dose level")
  expect_equal(result_mean_bio_fit$labels$y, "Biomarker level")

  # Check that plot has the expected structure
  expect_true(length(result_mean_bio_fit$layers) > 0)
  expect_equal(length(simSummary@dose_grid), length(simSummary@dose_grid))

  # Test combination with other plots
  result_multiple <- plot(simSummary, type = c("meanBiomarkerFit", "meanFit"))
  expect_s3_class(result_multiple, "gtable")
  expect_equal(dim(result_multiple), c(2, 1)) # Should have 2 rows, 1 column

  # Test additional dual-specific plot types
  result_n_obs <- plot(simSummary, type = "nObs")
  expect_doppel("plot_dualSimsSummary_nObs", result_n_obs)
  expect_s3_class(result_n_obs, "ggplot")

  result_dose_selected <- plot(simSummary, type = "doseSelected")
  expect_doppel("plot_dualSimsSummary_doseSelected", result_dose_selected)
  expect_s3_class(result_dose_selected, "ggplot")
})

## summary-PseudoSimulations ----

test_that("summary-PseudoSimulations works correctly", {
  # Create pseudo simulation
  emptydata <- Data(doseGrid = seq(25, 300, 25))

  model <- LogisticIndepBeta(
    binDLE = c(1.05, 1.8),
    DLEweights = c(3, 3),
    DLEdose = c(25, 300),
    data = emptydata
  )

  design <- TDDesign(
    model = model,
    nextBest = NextBestTD(prob_target_drt = 0.35, prob_target_eot = 0.3),
    stopping = StoppingMinPatients(nPatients = 6),
    increments = IncrementsRelative(intervals = 0, increments = 2),
    cohort_size = CohortSizeConst(size = 3),
    data = emptydata,
    startingDose = 25
  )

  myTruth <- probFunction(model, phi1 = -53.66584, phi2 = 10.50499)
  options <- McmcOptions(
    burnin = 10,
    step = 2,
    samples = 20,
    rng_kind = "Mersenne-Twister",
    rng_seed = 123
  )

  mySims <- simulate(
    design,
    args = NULL,
    truth = myTruth,
    nsim = 2,
    seed = 819,
    parallel = FALSE,
    mcmcOptions = options
  )

  result <- summary(mySims, truth = myTruth)
  expect_snapshot(result)
  expect_s4_class(result, "PseudoSimulationsSummary")
})

# show-PseudoSimulationsSummary ----

test_that("show-PseudoSimulationsSummary works correctly", {
  pseudo_sims <- .DefaultPseudoSimulations()
  pseudo_summary <- summary(
    pseudo_sims,
    truth = function(x) plogis(-3 + 0.05 * x),
    targetEndOfTrial = 0.3,
    targetDuringTrial = 0.35
  )

  # Test that show method produces output
  expect_output(show(pseudo_summary), "Summary of")
  expect_output(show(pseudo_summary), "simulations")
  expect_output(show(pseudo_summary), "Target probability of DLE")
  expect_output(show(pseudo_summary), "dose level corresponds")
  expect_output(show(pseudo_summary), "Number of patients overall")
  expect_output(show(pseudo_summary), "Number of patients treated above")

  # Test that it returns invisibly a data frame
  result <- capture.output(show_result <- show(pseudo_summary))
  expect_true(is.data.frame(show_result))
  expect_true(ncol(show_result) > 0)
  expect_snapshot(show(pseudo_summary))
})

test_that("show-PseudoSimulationsSummary produces expected output format", {
  pseudo_sims <- .DefaultPseudoSimulations()
  pseudo_summary <- summary(pseudo_sims, truth = function(x) {
    plogis(-3 + 0.05 * x)
  })

  # Capture the output for detailed testing
  output <- capture.output(show(pseudo_summary))

  # Test specific content patterns
  expect_true(any(grepl("Target probability.*end of a trial.*%", output)))
  expect_true(any(grepl("dose level corresponds.*target.*TDEOT", output)))
  expect_true(any(grepl("TDEOT at dose Grid", output)))
  expect_true(any(grepl("Target.*during a trial.*%", output)))
  expect_true(any(grepl("TDDT at dose Grid", output)))
  expect_true(any(grepl("Number of patients overall", output)))
  expect_true(any(grepl(
    "Number of patients treated above.*end of a trial",
    output
  )))
  expect_true(any(grepl(
    "Number of patients treated above.*during a trial",
    output
  )))
  expect_snapshot(show(pseudo_summary))
})

# plot-PseudoSimulationsSummary ----

test_that("plot-PseudoSimulationsSummary works correctly", {
  pseudo_sims <- .DefaultPseudoSimulations()
  pseudo_summary <- summary(pseudo_sims, truth = function(x) {
    plogis(-3 + 0.05 * x)
  })

  # Test default plot types
  result <- plot(pseudo_summary)
  expect_s3_class(result, "gtable")
  expect_equal(dim(result)[1], 2) # Should have 2 plots by default

  # Test individual plot types
  result_nobs <- plot(pseudo_summary, type = "nObs")
  expect_s3_class(result_nobs, "ggplot")
  expect_doppel("plot_pseudoSimsSummary_nObs", result_nobs)
  expect_equal(result_nobs$labels$x, "Number of patients in total")

  result_dose <- plot(pseudo_summary, type = "doseSelected")
  expect_s3_class(result_dose, "ggplot")
  expect_doppel("plot_pseudoSimsSummary_doseSelected", result_dose)
  expect_equal(result_dose$labels$x, "MTD estimate")

  result_prop <- plot(pseudo_summary, type = "propDLE")
  expect_s3_class(result_prop, "ggplot")
  expect_doppel("plot_pseudoSimsSummary_propDLE", result_prop)
  expect_equal(result_prop$labels$x, "Proportion of DLE [%]")

  result_target <- plot(pseudo_summary, type = "nAboveTargetEndOfTrial")
  expect_s3_class(result_target, "ggplot")
  expect_doppel(
    "plot_pseudoSimsSummary_nAboveTargetEndOfTrial",
    result_target
  )
  expect_equal(
    result_target$labels$x,
    "Number of patients above target"
  )

  result_mean <- plot(pseudo_summary, type = "meanFit")
  expect_s3_class(result_mean, "ggplot")
  expect_doppel("plot_pseudoSimsSummary_meanFit", result_mean)
  expect_true(grepl(
    "Probability of DLE [%]",
    result_mean$labels$y,
    fixed = TRUE
  ))
})

test_that("plot-PseudoSimulationsSummary handles multiple types", {
  pseudo_sims <- .DefaultPseudoSimulations()
  pseudo_summary <- summary(pseudo_sims, truth = function(x) {
    plogis(-3 + 0.05 * x)
  })

  # Test multiple specific types
  result_multi <- plot(pseudo_summary, type = c("nObs", "doseSelected"))
  expect_s3_class(result_multi, "gtable")
  expect_equal(dim(result_multi)[1], 2) # Should have 2 plots

  # Test with all types explicitly
  all_types <- c(
    "nObs",
    "doseSelected",
    "propDLE",
    "nAboveTargetEndOfTrial",
    "meanFit"
  )
  result_all <- plot(pseudo_summary, type = all_types)
  expect_s3_class(result_all, "gtable")
  expect_equal(dim(result_all)[1], 2) # Should have 2 plots
})

test_that("plot-PseudoSimulationsSummary fails gracefully with bad input", {
  pseudo_sims <- .DefaultPseudoSimulations()
  pseudo_summary <- summary(pseudo_sims, truth = function(x) {
    plogis(-3 + 0.05 * x)
  })

  expect_error(
    plot(pseudo_summary, type = "invalid_type"),
    "should be one of"
  )
  expect_error(plot(pseudo_summary, type = character(0)), "must be of length")
})

# plot-PseudoDualSimulations ----

test_that("plot-PseudoDualSimulations works correctly", {
  pseudo_dual_sims <- .DefaultPseudoDualSimulations()

  # Test default plot types
  result <- plot(pseudo_dual_sims)
  expect_s3_class(result, "gtable")

  # Test individual plot types
  result_trajectory <- plot(pseudo_dual_sims, type = "trajectory")
  expect_s3_class(result_trajectory, "ggplot")
  expect_doppel("plot_pseudoDualSims_trajectory", result_trajectory)
  expect_equal(result_trajectory$labels$x, "Patient")
  expect_equal(result_trajectory$labels$y, "Dose Level")

  result_doses <- plot(pseudo_dual_sims, type = "dosesTried")
  expect_s3_class(result_doses, "ggplot")
  expect_doppel("plot_pseudoDualSims_dosesTried", result_doses)
  expect_equal(result_doses$labels$x, "Dose level")
  expect_equal(result_doses$labels$y, "Average proportion [%]")

  result_sigma2 <- plot(pseudo_dual_sims, type = "sigma2")
  expect_s3_class(result_sigma2, "ggplot")
  expect_doppel("plot_pseudoDualSims_sigma2", result_sigma2)
  expect_true(grepl(
    "Efficacy variance estimates",
    result_sigma2$labels$y,
    ignore.case = TRUE
  ))
})

test_that("plot-PseudoDualSimulations handles type combinations", {
  pseudo_dual_sims <- .DefaultPseudoDualSimulations()

  # Test multiple types
  result_multi <- plot(pseudo_dual_sims, type = c("trajectory", "sigma2"))
  expect_s3_class(result_multi, "gtable")

  # Test all available types
  result_all <- plot(
    pseudo_dual_sims,
    type = c("trajectory", "dosesTried", "sigma2")
  )
  expect_s3_class(result_all, "gtable")
})

test_that("plot-PseudoDualSimulations fails gracefully with bad input", {
  pseudo_dual_sims <- .DefaultPseudoDualSimulations()

  expect_error(
    plot(pseudo_dual_sims, type = "invalid_type"),
    "should be one of"
  )
  expect_error(
    plot(pseudo_dual_sims, type = character(0)),
    "must be of length"
  )
})

# plot-PseudoDualFlexiSimulations ----

test_that("plot-PseudoDualFlexiSimulations works correctly", {
  pseudo_dual_flexi_sims <- .DefaultPseudoDualFlexiSimulations()

  # Test default plot types
  result <- plot(pseudo_dual_flexi_sims)
  expect_s3_class(result, "gtable")

  # Test individual plot types
  result_trajectory <- plot(pseudo_dual_flexi_sims, type = "trajectory")
  expect_s3_class(result_trajectory, "ggplot")
  expect_doppel("plot_pseudoDualFlexiSims_trajectory", result_trajectory)
  expect_equal(result_trajectory$labels$x, "Patient")
  expect_equal(result_trajectory$labels$y, "Dose Level")

  result_doses <- plot(pseudo_dual_flexi_sims, type = "dosesTried")
  expect_s3_class(result_doses, "ggplot")
  expect_doppel("plot_pseudoDualFlexiSims_dosesTried", result_doses)
  expect_equal(result_doses$labels$x, "Dose level")
  expect_equal(result_doses$labels$y, "Average proportion [%]")

  result_sigma2 <- plot(pseudo_dual_flexi_sims, type = "sigma2")
  expect_s3_class(result_sigma2, "ggplot")
  expect_doppel("plot_pseudoDualFlexiSims_sigma2", result_sigma2)
  expect_true(grepl(
    "Efficacy variance estimates",
    result_sigma2$labels$y,
    ignore.case = TRUE
  ))

  result_sigma2beta_w <- plot(pseudo_dual_flexi_sims, type = "sigma2betaW")
  expect_s3_class(result_sigma2beta_w, "ggplot")
  expect_doppel("plot_pseudoDualFlexiSims_sigma2betaW", result_sigma2beta_w)
  expect_true(grepl(
    "Random walk model variance estimates",
    result_sigma2beta_w$labels$y,
    ignore.case = TRUE
  ))
})

# summary-PseudoDualSimulations ----

test_that("summary-PseudoDualSimulations works correctly", {
  pseudo_dual_sims <- .DefaultPseudoDualSimulations()

  # Test basic summary call
  result <- summary(
    pseudo_dual_sims,
    trueDLE = function(x) plogis(-3 + 0.05 * x),
    trueEff = function(x) 0.2 + 0.004 * x
  )

  expect_s4_class(result, "PseudoDualSimulationsSummary")
  expect_snapshot(result)

  # Test with custom target values
  result_custom <- summary(
    pseudo_dual_sims,
    trueDLE = function(x) plogis(-3 + 0.05 * x),
    trueEff = function(x) 0.2 + 0.004 * x,
    targetEndOfTrial = 0.25,
    targetDuringTrial = 0.30
  )

  expect_s4_class(result_custom, "PseudoDualSimulationsSummary")
  expect_equal(result_custom@target_end_of_trial, 0.25)
  expect_snapshot(result_custom)
})

test_that("summary-PseudoDualSimulations produces expected structure", {
  pseudo_dual_sims <- .DefaultPseudoDualSimulations()

  result <- summary(
    pseudo_dual_sims,
    trueDLE = function(x) plogis(-3 + 0.05 * x),
    trueEff = function(x) 0.2 + 0.004 * x
  )

  # Test inheritance from parent class
  expect_true(is(result, "PseudoDualSimulationsSummary"))
  expect_true(is(result, "PseudoSimulationsSummary"))
  expect_snapshot(result)

  # Test that dual-specific slots exist
  expect_true("target_gstar" %in% slotNames(result))
  expect_true("target_gstar_at_dose_grid" %in% slotNames(result))
  expect_true("gstar_summary" %in% slotNames(result))
  expect_true("eff_fit_at_dose_most_selected" %in% slotNames(result))
  expect_true("mean_eff_fit" %in% slotNames(result))
})

# summary-PseudoDualFlexiSimulations ----

test_that("summary-PseudoDualFlexiSimulations works correctly", {
  pseudo_dual_flexi_sims <- .DefaultPseudoDualFlexiSimulations()

  # Test basic summary call
  result <- summary(
    pseudo_dual_flexi_sims,
    trueDLE = function(x) plogis(-3 + 0.05 * x),
    trueEff = function(x) 0.2 + 0.004 * x
  )

  expect_s4_class(result, "PseudoDualSimulationsSummary")
  expect_snapshot(result)

  # Test with custom target values
  result_custom <- summary(
    pseudo_dual_flexi_sims,
    trueDLE = function(x) plogis(-3 + 0.05 * x),
    trueEff = function(x) 0.2 + 0.004 * x,
    targetEndOfTrial = 0.25,
    targetDuringTrial = 0.30
  )

  expect_s4_class(result_custom, "PseudoDualSimulationsSummary")
  expect_equal(result_custom@target_end_of_trial, 0.25)
  expect_snapshot(result_custom)
})

# show-PseudoDualSimulationsSummary ----

test_that("show-PseudoDualSimulationsSummary works correctly", {
  pseudo_dual_sims <- .DefaultPseudoDualSimulations()
  pseudo_dual_summary <- summary(
    pseudo_dual_sims,
    trueDLE = function(x) plogis(-3 + 0.05 * x),
    trueEff = function(x) 0.2 + 0.004 * x
  )

  # Test that show method produces output
  expect_snapshot(show(pseudo_dual_summary))
  expect_output(show(pseudo_dual_summary), "Target Gstar")
  expect_output(show(pseudo_dual_summary), "maximum gain value")
  expect_output(show(pseudo_dual_summary), "Target Gstar at dose Grid")

  # Test that it calls parent method
  expect_output(show(pseudo_dual_summary), "Summary of")
  expect_output(show(pseudo_dual_summary), "simulations")

  # Test that it returns a data frame invisibly
  result_output <- capture.output(show_result <- show(pseudo_dual_summary))
  expect_true(is.data.frame(show_result))
})

test_that("show-PseudoDualSimulationsSummary includes dual-specific content", {
  pseudo_dual_sims <- .DefaultPseudoDualSimulations()
  pseudo_dual_summary <- summary(
    pseudo_dual_sims,
    trueDLE = function(x) plogis(-3 + 0.05 * x),
    trueEff = function(x) 0.2 + 0.004 * x
  )

  output <- capture.output(show(pseudo_dual_summary))
  expect_snapshot(show(pseudo_dual_summary))

  # Test dual-specific content
  expect_true(any(grepl("Target Gstar.*maximum gain value", output)))
  expect_true(any(grepl("Target Gstar at dose Grid", output)))

  # Test that parent content is also included
  expect_true(any(grepl("Target probability of DLE", output)))
  expect_true(any(grepl("Number of patients", output)))
})

# plot-PseudoDualSimulationsSummary ----

test_that("plot-PseudoDualSimulationsSummary works correctly", {
  pseudo_dual_sims <- .DefaultPseudoDualSimulations()
  pseudo_dual_summary <- summary(
    pseudo_dual_sims,
    trueDLE = function(x) plogis(-3 + 0.05 * x),
    trueEff = function(x) 0.2 + 0.004 * x
  )

  # Test default behavior
  result <- plot(pseudo_dual_summary)
  expect_s3_class(result, "gtable")

  # Test dual-specific plot type
  result_eff <- plot(pseudo_dual_summary, type = "meanEffFit")
  expect_s3_class(result_eff, "ggplot")
  expect_doppel("plot_pseudoDualSimsSummary_meanEffFit", result_eff)
  expect_true(
    grepl("efficacy", result_eff$labels$y, ignore.case = TRUE) ||
      grepl("eff", result_eff$labels$y, ignore.case = TRUE)
  )

  # Test combination with parent types
  result_combo <- plot(pseudo_dual_summary, type = c("nObs", "meanEffFit"))
  expect_s3_class(result_combo, "gtable")
  expect_equal(dim(result_combo)[1], 2)

  # Test all available types
  all_types <- c(
    "nObs",
    "doseSelected",
    "propDLE",
    "nAboveTargetEndOfTrial",
    "meanFit",
    "meanEffFit"
  )
  result_all <- plot(pseudo_dual_summary, type = all_types)
  expect_s3_class(result_all, "gtable")
  expect_equal(dim(result_all)[1], 2)
})

test_that("plot-PseudoDualSimulationsSummary handles edge cases", {
  pseudo_dual_sims <- .DefaultPseudoDualSimulations()
  pseudo_dual_summary <- summary(
    pseudo_dual_sims,
    trueDLE = function(x) plogis(-3 + 0.05 * x),
    trueEff = function(x) 0.2 + 0.004 * x
  )

  # Test error handling
  expect_error(
    plot(pseudo_dual_summary, type = "invalid_type"),
    "should be one of"
  )
  expect_error(
    plot(pseudo_dual_summary, type = character(0)),
    "must be of length"
  )

  # Test single dual-specific type
  result_single <- plot(pseudo_dual_summary, type = "meanEffFit")
  expect_s3_class(result_single, "ggplot")
  expect_doppel("plot_pseudoDualSimsSummary_meanEffFit_single", result_single)
})