tests/testthat/test-upsample.R

In keyes-timothy/tidytof: Analyze High-dimensional Cytometry Data Using Tidy Data Principles

# libraries
library(tidytof)
library(dplyr)
library(testthat)

# set up
data(ddpr_data)

reference_tibble <-
    ddpr_data |>
    slice_sample(prop = 0.1) |>
    mutate(my_cluster = sample(x = c("a", "b", "c"), size = n(), replace = TRUE))

# tof_upsample_distance --------------------------------------------------------

test_that("upsample result is a tibble with a single character vector column of correct length", {
    result <-
        ddpr_data |>
        tof_upsample_distance(
            reference_tibble = reference_tibble,
            reference_cluster_col = my_cluster,
            upsample_cols = c(cd34, cd45, cd19),
            return_distances = FALSE
        )

    expect_equal(nrow(ddpr_data), nrow(result))
    expect_true(is.character(result$.upsample_cluster))
    expect_equal(ncol(result), 1L)
})

test_that("return_distances argument works", {
    result <-
        ddpr_data |>
        tof_upsample_distance(
            reference_tibble = reference_tibble,
            reference_cluster_col = my_cluster,
            upsample_cols = c(cd34, cd45, cd19),
            return_distances = TRUE
        )

    expect_equal(ncol(result), 4L)
})

test_that("output columns are named correctly", {
    result_mah <-
        ddpr_data |>
        tof_upsample_distance(
            reference_tibble = reference_tibble,
            reference_cluster_col = my_cluster,
            upsample_cols = c(cd34, cd45, cd19),
            return_distances = TRUE
        )

    result_cosine <-
        ddpr_data |>
        tof_upsample_distance(
            reference_tibble = reference_tibble,
            reference_cluster_col = my_cluster,
            upsample_cols = c(cd34, cd45, cd19),
            distance_function = "cosine",
            return_distances = TRUE,
            parallel_cols = sample_name
        )

    result_pearson <-
        ddpr_data |>
        tof_upsample_distance(
            reference_tibble = reference_tibble,
            reference_cluster_col = my_cluster,
            upsample_cols = c(cd34, cd45, cd19),
            distance_function = "pearson",
            return_distances = TRUE
        )

    # all column names should start with "."
    expect_true(all(grepl("^\\.", colnames(result_mah))))
    expect_true(all(grepl("^\\.", colnames(result_cosine))))
    expect_true(all(grepl("^\\.", colnames(result_pearson))))

    # column names for distances should start with ".{distance_function}"
    expect_equal(sum(grepl("\\.mahalanobis", colnames(result_mah))), 3L)
    expect_equal(sum(grepl("\\.cosine", colnames(result_cosine))), 3L)
    expect_equal(sum(grepl("\\.pearson", colnames(result_pearson))), 3L)
})

# tof_upsample_neighbor --------------------------------------------------------

test_that("upsample result is a tibble with a single character vector column of correct length", {
    result <-
        ddpr_data |>
        tof_upsample_neighbor(
            reference_tibble = reference_tibble,
            reference_cluster_col = my_cluster,
            upsample_cols = c(cd34, cd45, cd19)
        )

    expect_equal(nrow(ddpr_data), nrow(result))
    expect_true(is.character(result$.upsample_cluster))
    expect_equal(ncol(result), 1L)
})

test_that("output columns are named correctly", {
    result <-
        ddpr_data |>
        tof_upsample_neighbor(
            reference_tibble = reference_tibble,
            reference_cluster_col = my_cluster,
            upsample_cols = c(cd34, cd45, cd19)
        )

    expect_identical(colnames(result), ".upsample_cluster")
})

# tof_upsample -----------------------------------------------------------------

test_that("tof_upsample and tof_upsample_distance results are the same", {
    result_1 <-
        ddpr_data |>
        tof_upsample_distance(
            reference_tibble = reference_tibble,
            reference_cluster_col = my_cluster,
            upsample_cols = c(cd34, cd45, cd19),
            return_distances = TRUE
        )

    result_2 <-
        ddpr_data |>
        tof_upsample(
            reference_tibble = reference_tibble,
            reference_cluster_col = my_cluster,
            upsample_cols = c(cd34, cd45, cd19),
            return_distances = TRUE,
            augment = FALSE,
            method = "distance"
        )

    expect_equal(result_1, result_2)
})

test_that("tof_upsample and tof_upsample_neighbor results are the same", {
    result_1 <-
        ddpr_data |>
        tof_upsample_neighbor(
            reference_tibble = reference_tibble,
            reference_cluster_col = my_cluster,
            upsample_cols = c(cd34, cd45, cd19),
        )

    result_2 <-
        ddpr_data |>
        tof_upsample(
            reference_tibble = reference_tibble,
            reference_cluster_col = my_cluster,
            upsample_cols = c(cd34, cd45, cd19),
            augment = FALSE,
            method = "neighbor"
        )

    expect_equal(result_1, result_2)
})