Update slice_sample in dplyr_methods.R

[noriakis]

Hello, this PR clears the warning in slice_sample. I would be grateful if you could review the code.
Related to: stemangiola/tidySingleCellExperiment#90

> data(pbmc_small)
> pbmc_small |> slice_sample(n=0)
Error: No cells found

[stemangiola]

Amazing!

Usually when we add a functionality, or fix an error/warning, we add a unit test, to test the changes.

For exmple if we solve an error, we would

• first add a unit test (that would fail) to test that the error happens
• we then fix the error
• rerun the unit test expecting no error
• we commit the fix and the unit test

[noriakis]

Thank you for your guidance! In this case, I added the unit test for slice_sample() expecting the error (expect_error). Would it be okay?

Edit: I apologize for the recurrent commit, but I thought I should have included expect_no_warning as well.

[stemangiola]

Hello @wvictor14 @noriakis In the Seurat master, the slice functions produce errors in the tests. Could you please have a look?

Thanks!

[noriakis]

Hello @stemangiola @wvictor14,

expect_equal(
    pbmc_small |> as_tibble() |> arrange(nFeature_RNA) |> head(n = 5) %>% pull(.cell),
    pbmc_small |> slice_min(nFeature_RNA, n = 5) |> colnames()
)

The errors seem to come from above tests in slice_min and slice_max, where the subset.Seurat by cells returns the object that is preserving the original order of the cells. The returned row numbers of cells inside the function are correct, so one option would be to adding sort() to both vectors which removes the errors.

expect_equal(
    pbmc_small |> as_tibble() |> arrange(nFeature_RNA) |> head(n = 5) %>% pull(.cell) |> sort(),
    pbmc_small |> slice_min(nFeature_RNA, n = 5) |> colnames() |> sort()
)

The arrange() function is currently deprecated, so maybe we could replace the test?

[stemangiola]

The arrange() function is currently deprecated, so maybe we could replace the test?

Agree, if you had the chance to do it, would be amazing.