Tests #76
Tests #76
Conversation
…ient_representation into 68-inconsistent-interface
VladimirShitov
left a comment
VladimirShitov
left a comment
There was a problem hiding this comment.
Thanks! Looks great overall. Please, resolve small comments I left, and use fixtures wherever possible
|
|
||
|
|
||
| def test_distances_significance( | ||
| def distances_significance_test( |
There was a problem hiding this comment.
Why renaming? I also prefer starting tests with test_
There was a problem hiding this comment.
Ah, because this is not the test. Thanks for changing it then, makes sense!
|
|
||
|
|
||
| def test_proportions(target, groups): | ||
| def statistical_test_proportions(target, groups): |
There was a problem hiding this comment.
Let's do it like that:
| def statistical_test_proportions(target, groups): | |
| def composition_differences_test(target, groups): |
|
|
||
|
|
||
| @pytest.fixture | ||
| def toy_dataframe(): |
There was a problem hiding this comment.
Could you call it toy_metadata if that's what it it? Not clear what an abstract data frame means
| """Creates a toy distance matrix and conditions array for testing.""" | ||
| distances = squareform(pdist(np.array([[0, 1], [1, 1], [2, 2], [3, 3]]))) | ||
| conditions = pd.Series(["control", "control", "case", "case"]) | ||
| # conditions = np.array(["control", "control", "case", "case"]) |
There was a problem hiding this comment.
Please remove commented code :)
|
|
||
| assert isinstance(result, pd.DataFrame), "The result should be a DataFrame." | ||
| assert result.shape[0] == 2, "The result should contain metadata for 2 samples." | ||
| assert "cell_type" in result.columns, "The result should contain the cell_type column." |
There was a problem hiding this comment.
In a normal scenario, that's not what we want. Cell type is cell-level information, and we want to extract donor-level metadata. The test checks the correct thing, but could you change the column to something else to be less confusing?
| ), "There should be a unique color for each unique value in the column." | ||
|
|
||
|
|
||
| def test_describe_metadata(toy_dataframe, capsys): |
There was a problem hiding this comment.
Haha, that's a bit too much. The tested function is just for convenience. But thanks for testing it as well
|
|
||
| def test_calculate_average_without_nans(): | ||
| # no nan | ||
| array = np.array( |
There was a problem hiding this comment.
I'd suggest using fixtures and give them readable names :) E.g. integer_matrix (can also be used it testing count data), float_matrix_with_nans, and so on
| y_true = np.array([0, 1, 1, 0]) | ||
| y_pred = np.array([0, 1, 1, 0]) | ||
|
|
||
| result = evaluate_prediction(y_true, y_pred, task="classification") |
There was a problem hiding this comment.
Could you test other tasks as well?
| 2, | ||
| 2, | ||
| ), "Filtered distances should have shape (2, 2) after removing missing values." | ||
| assert filtered_target.shape[0] == 2, "Filtered target should have 2 values after removing missing values." |
There was a problem hiding this comment.
You can also check the exact values as you know what to expect in this example
|
|
||
|
|
||
| def test_select_random_subset(): | ||
| distances = np.array([[0, 1, 2], [1, 0, 3], [2, 3, 0]]) |
There was a problem hiding this comment.
Use fixtures wherever possible :) It allows to easily extend tests with new data
No description provided.