tests/testthat/test_umapscan.R

In juba/umapscan: UMAP and DBSCAN Pragmatic and Supervised clustering

context("umapscan class functions")

iris_num <- iris[, c("Sepal.Length", "Sepal.Width", "Petal.Length", "Petal.Width")]
iris_sup <- iris[, "Species", drop=FALSE]

test_that("new_umapscan with wrong arguments throws error", {
  expect_error(new_umapscan(d = 1:10))
  expect_error(new_umapscan(d = iris_num, data_sup = 1:10))
  expect_error(new_umapscan(d = iris_num, data_sup = iris_sup[-1, ]))
})

test_that("new_umapscan with matrix argument", {
  d <- data.frame(x = 1:5, y = 1:5, z = 1:5)
  m <- matrix(c(1:5, 1:5, 1:5), ncol = 3, dimnames = list(1:5, c("x", "y", "z")))
  usd <- new_umapscan(d, seed = 12, n_neighbors = 3, min_dist = 0.1)
  usm <- new_umapscan(m, seed = 12, n_neighbors = 3, min_dist = 0.1)
  expect_equal(usd$umap, usm$umap)
})

test_that("new_umapscan results are ok", {
  us <- new_umapscan(iris_num, data_sup = iris_sup, seed = 1337, scale = FALSE)
  expect_equal(us$data, iris_num)
  expect_equal(us$data_sup, iris_sup)
  expect_equal(dim(us$cluster), c(0, 6))
  expect_equal(dim(us$umap), c(nrow(iris_num), 2))
  set.seed(1337)
  umap <- uwot::umap(iris_num, approx_pow = TRUE)
  expect_equal(tibble::tibble(.umap_x = umap[,1], .umap_y = umap[,2]), us$umap)
})

test_that("new_umapscan results with scale=TRUE are ok", {
  us <- new_umapscan(iris_num, data_sup = iris_sup, seed = 24312, scale = TRUE)
  set.seed(24312)
  umap <- uwot::umap(dplyr::mutate_all(iris_num, base::scale),  approx_pow = TRUE)
  expect_equal(tibble::tibble(.umap_x = umap[,1], .umap_y = umap[,2]), us$umap)
})

test_that("new_umapscan results are reproducible", {
  us <- new_umapscan(iris_num, seed = 24312, scale = TRUE)
  # saveRDS(us$umap %>% dplyr::slice(1:50) %>% data.frame, "tests/values/umapscan1.rds")
  expect_equal(
    us$umap %>% dplyr::slice(1:50) %>% data.frame,
    readRDS("../values/umapscan1.rds")
  )
})

test_that("set.seed is ok", {
  set.seed(556677)
  x <- runif(1)
  expect_equal(x, 0.05051134, tolerance = 0.00000001)
  y <- rnorm(1)
  expect_equal(y, 1.970592, tolerance = 0.000001)
  z <- runif(1)
  expect_equal(z, 0.3479608, tolerance = 0.0000001)
})

test_that("base umapscan with seed is ok", {
  set.seed(8222)
  umap <- uwot::umap(USArrests, approx_pow = TRUE)
  # saveRDS(umap, "tests/values/umapscan2.rds")
  expect_equivalent(
    umap,
    readRDS("../values/umapscan2.rds")
  )
  set.seed(8223)
  umap_pca <- uwot::umap(USArrests, init = "pca", approx_pow = TRUE)
  # saveRDS(umap_pca, "tests/values/umapscan3.rds")
  expect_equivalent(
    umap_pca,
    readRDS("../values/umapscan3.rds")
  )
})