tests/testthat/test-metaclustering.R

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

library(dplyr)
library(tidytof)
library(testthat)

data(phenograph_data)

# setup
clust_data <-
    phenograph_data |>
    tof_preprocess() |>
    tof_cluster(method = "kmeans")

# tof_metacluster_hierarchical -------------------------------------------------
test_that("metaclustering result is the right shape", {
    result <-
        clust_data |>
        tof_metacluster_hierarchical(
            cluster_col = .kmeans_cluster
        )

    expect_equal(nrow(result), nrow(clust_data))
    expect_equal(ncol(result), 1L)
})

test_that("metaclustering result has the right name", {
    result <-
        clust_data |>
        tof_metacluster_hierarchical(
            cluster_col = .kmeans_cluster
        )

    expect_equal(colnames(result), ".hierarchical_metacluster")
})



# tof_metacluster_kmeans -------------------------------------------------------
test_that("metaclustering result is the right shape", {
    result <-
        clust_data |>
        tof_metacluster_kmeans(
            cluster_col = .kmeans_cluster
        )

    expect_equal(nrow(result), nrow(clust_data))
    expect_equal(ncol(result), 1L)
})

test_that("metaclustering result has the right name", {
    result <-
        clust_data |>
        tof_metacluster_kmeans(
            cluster_col = .kmeans_cluster
        )

    expect_equal(colnames(result), ".kmeans_metacluster")
})


# tof_metacluster_phenograph ---------------------------------------------------
test_that("metaclustering result is the right shape", {
    result <-
        clust_data |>
        tof_metacluster_phenograph(
            cluster_col = .kmeans_cluster
        )

    expect_equal(nrow(result), nrow(clust_data))
    expect_equal(ncol(result), 1L)
})

test_that("metaclustering result has the right name", {
    result <-
        clust_data |>
        tof_metacluster_phenograph(
            cluster_col = .kmeans_cluster
        )

    expect_equal(colnames(result), ".phenograph_metacluster")
})

test_that("metaclustering result works with various kinds of tidyselection", {
    result_1 <-
        clust_data |>
        tof_metacluster_phenograph(
            cluster_col = .kmeans_cluster,
            metacluster_cols = contains("cd")
        )

    result_2 <-
        clust_data |>
        tof_metacluster_phenograph(
            cluster_col = .kmeans_cluster,
            metacluster_cols = c(contains("cd"), -cd45)
        )

    result_3 <-
        clust_data |>
        tof_metacluster_phenograph(
            cluster_col = .kmeans_cluster,
            metacluster_cols = c(cd45)
        )

    result_4 <-
        clust_data |>
        tof_metacluster_phenograph(
            cluster_col = .kmeans_cluster,
            metacluster_cols = c(cd45, -sample_name)
        )

    expect_equal(nrow(result_1), nrow(clust_data))
    expect_equal(ncol(result_1), 1L)
    expect_equal(nrow(result_2), nrow(clust_data))
    expect_equal(ncol(result_2), 1L)
    expect_equal(nrow(result_3), nrow(clust_data))
    expect_equal(ncol(result_3), 1L)
    expect_equal(nrow(result_4), nrow(clust_data))
    expect_equal(ncol(result_4), 1L)
})

# tof_metacluster_consensus ----------------------------------------------------
test_that("metaclustering result is the right shape", {
    result <-
        clust_data |>
        tof_metacluster_consensus(
            cluster_col = .kmeans_cluster,
            clustering_algorithm = "kmeans"
        )

    result_2 <-
        clust_data |>
        tof_metacluster_consensus(
            cluster_col = .kmeans_cluster,
            clustering_algorithm = "pam"
        )

    result_3 <-
        clust_data |>
        tof_metacluster_consensus(
            cluster_col = .kmeans_cluster,
            clustering_algorithm = "hierarchical"
        )

    result_4 <-
        clust_data |>
        tof_metacluster_consensus(
            cluster_col = .kmeans_cluster,
            clustering_algorithm = "hierarchical",
            distance_function = "pearson"
        )

    expect_equal(nrow(result), nrow(clust_data))
    expect_equal(ncol(result), 1L)

    expect_equal(nrow(result_2), nrow(clust_data))
    expect_equal(ncol(result_2), 1L)

    expect_equal(nrow(result_3), nrow(clust_data))
    expect_equal(ncol(result_3), 1L)

    expect_equal(nrow(result_4), nrow(clust_data))
    expect_equal(ncol(result_4), 1L)
})

test_that("metaclustering result has the right name", {
    result <-
        clust_data |>
        tof_metacluster_consensus(
            cluster_col = .kmeans_cluster
        )

    expect_equal(colnames(result), ".consensus_metacluster")
})


# tof_metacluster_flowsom ------------------------------------------------------
test_that("metaclustering result is the right shape", {
    result <-
        clust_data |>
        tof_metacluster_flowsom(
            cluster_col = .kmeans_cluster,
            clustering_algorithm = "consensus"
        )

    result_2 <-
        clust_data |>
        tof_metacluster_flowsom(
            cluster_col = .kmeans_cluster,
            clustering_algorithm = "hierarchical"
        )

    result_3 <-
        clust_data |>
        tof_metacluster_flowsom(
            cluster_col = .kmeans_cluster,
            clustering_algorithm = "som"
        )

    result_4 <-
        clust_data |>
        tof_metacluster_flowsom(
            cluster_col = .kmeans_cluster,
            clustering_algorithm = "kmeans"
        )

    expect_equal(nrow(result), nrow(clust_data))
    expect_equal(ncol(result), 1L)

    expect_equal(nrow(result_2), nrow(clust_data))
    expect_equal(ncol(result_2), 1L)

    expect_equal(nrow(result_3), nrow(clust_data))
    expect_equal(ncol(result_3), 1L)

    expect_equal(nrow(result_4), nrow(clust_data))
    expect_equal(ncol(result_4), 1L)
})

test_that("metaclustering result has the right name", {
    result <-
        clust_data |>
        tof_metacluster_flowsom(
            cluster_col = .kmeans_cluster
        )

    expect_equal(colnames(result), ".flowsom_metacluster")
})

# tof_metacluster --------------------------------------------------------------

test_that("hclust tof_metacluster results are identical to subroutine", {
    hclust_sub <-
        clust_data |>
        tof_metacluster_hierarchical(
            cluster_col = .kmeans_cluster
        )

    hclust_wrap <-
        clust_data |>
        tof_metacluster(
            cluster_col = .kmeans_cluster,
            augment = FALSE,
            method = "hierarchical"
        )

    expect_equal(hclust_sub, hclust_wrap)
})

test_that("kmeans tof_metacluster results are identical to subroutine", {
    set.seed(2020)
    kmeans_sub <-
        clust_data |>
        tof_metacluster_kmeans(
            cluster_col = .kmeans_cluster
        )

    set.seed(2020)
    kmeans_wrap <-
        clust_data |>
        tof_metacluster(
            cluster_col = .kmeans_cluster,
            augment = FALSE,
            method = "kmeans"
        )

    expect_equal(kmeans_sub, kmeans_wrap)
})

test_that("phenograph tof_metacluster results are identical to subroutine", {
    set.seed(2020)
    pheno_sub <-
        clust_data |>
        tof_metacluster_phenograph(
            cluster_col = .kmeans_cluster
        )

    set.seed(2020)
    pheno_wrap <-
        clust_data |>
        tof_metacluster(
            cluster_col = .kmeans_cluster,
            augment = FALSE,
            method = "phenograph"
        )

    expect_equal(pheno_sub, pheno_wrap)
})

test_that("consensus tof_metacluster results are identical to subroutine", {
    ccp_sub <-
        clust_data |>
        tof_metacluster_consensus(
            cluster_col = .kmeans_cluster,
            seed = 2020L
        )

    ccp_wrap <-
        clust_data |>
        tof_metacluster(
            cluster_col = .kmeans_cluster,
            seed = 2020L,
            augment = FALSE,
            method = "consensus"
        )

    expect_equal(ccp_sub, ccp_wrap)
})

test_that("flowsom tof_metacluster results are identical to subroutine", {
    set.seed(2020)
    flow_sub <-
        clust_data |>
        tof_metacluster_flowsom(
            cluster_col = .kmeans_cluster,
        )

    set.seed(2020)
    flow_wrap <-
        clust_data |>
        tof_metacluster(
            cluster_col = .kmeans_cluster,
            augment = FALSE,
            method = "flowsom"
        )

    expect_equal(flow_sub, flow_wrap)
})