tests/testthat/test-find_markers.R

In cole-trapnell-lab/monocle3: Clustering, Differential Expression, and Trajectory Analysis for Single-Cell RNA-Seq

context("test-find_markers")

chk <- Sys.getenv("_R_CHECK_LIMIT_CORES_", "")

# top_markers(..., cores=8) earlier but devtools::check() fails because cores is too large.
# see https://stackoverflow.com/questions/50571325/r-cran-check-fail-when-using-parallel-functions
# Rather than use parallel::detectCores(), which detects threads rather than cores, may
# may call for too many cores under some circumstances?
if (nzchar(chk) && chk == "TRUE") {
    # use 2 cores in CRAN/Travis/AppVeyor
    num_cores <- 2L
} else {
    # use all cores in devtools::test()
    num_cores <- parallel::detectCores()
}

library(dplyr)
set.seed(1000)
cds <- load_a549()
assigned_type_marker_test_res = top_markers(cds,
                                            group_cells_by="top_oligo",
                                            reference_cells=1000,
                                            cores=num_cores)

garnett_markers <- assigned_type_marker_test_res %>%
  filter(marker_test_p_value < 0.01 & specificity >= 0.5) %>%
  group_by(cell_group) %>%
  top_n(5, marker_score)


test_that("generate_garnett_marker_file_works", {
  temp <- tempdir()
  generate_garnett_marker_file(garnett_markers, file=paste0(temp,"/marker_file.txt"))
  marker <- readLines(paste0(temp,"/marker_file.txt"))
  expect_equal(length(marker), 42)
  expect_equal(marker[1], "> Cell type Dex_0_AD04")

  garnett_markers$cell_group[1] <- "test_(_:_)_#_,_>"
  expect_warning(generate_garnett_marker_file(garnett_markers, file=paste0(temp,"/marker_file.txt")))
  marker <- readLines(paste0(temp,"/marker_file.txt"))
  expect_equal(length(marker), 45)
  expect_equal(marker[1], "> Cell type test_._._._._._.")
  unlink(temp)
})