-
Notifications
You must be signed in to change notification settings - Fork 5
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Signed-off-by: Cameron Smith <[email protected]>
- Loading branch information
1 parent
c28a598
commit 1480351
Showing
1 changed file
with
134 additions
and
2 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -1,7 +1,139 @@ | ||
"""Tests for `pyrovelocity.analysis.cytotrace` module.""" | ||
|
||
|
||
import numpy as np | ||
import pandas as pd | ||
import pytest | ||
from anndata import AnnData | ||
from numpy.testing import assert_array_almost_equal | ||
from scipy.sparse import csr_matrix | ||
|
||
from pyrovelocity.analysis import cytotrace | ||
|
||
|
||
def test_load_cytotrace(): | ||
pass | ||
print(cytotrace.__file__) | ||
|
||
|
||
@pytest.fixture | ||
def small_anndata(): | ||
X = csr_matrix([[1, 2, 3], [4, 5, 6], [7, 8, 9]]) | ||
obs = pd.DataFrame(index=["cell1", "cell2", "cell3"]) | ||
var = pd.DataFrame(index=["gene1", "gene2", "gene3"]) | ||
adata = AnnData(X, obs=obs, var=var) | ||
adata.layers["raw"] = X | ||
return adata | ||
|
||
|
||
def test_compute_similarity2(): | ||
O = np.array([[1, 2, 3], [4, 5, 6]]) | ||
P = np.array([[1, 2], [3, 4]]) | ||
result = cytotrace.compute_similarity2(O, P) | ||
assert result.shape == (2, 3) | ||
assert np.allclose(result.T, np.corrcoef(O.T, P)[:3, 3:], atol=1e-5) | ||
|
||
|
||
def test_compute_similarity1(): | ||
A = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]) | ||
result = cytotrace.compute_similarity1(A) | ||
assert result.shape == (3, 3) | ||
assert np.allclose(result, np.corrcoef(A.T)) | ||
|
||
|
||
def test_compute_gcs(): | ||
mat = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]) | ||
count = np.array([2, 3, 3]) | ||
result = cytotrace.compute_gcs(mat, count, top_n_genes=2) | ||
assert result.shape == (3,) | ||
|
||
|
||
def test_threshold_and_normalize_similarity_matrix(): | ||
sim = np.array( | ||
[ | ||
[1.0, 0.8, 0.3, 0.1], | ||
[0.8, 1.0, 0.5, 0.2], | ||
[0.3, 0.5, 1.0, 0.7], | ||
[0.1, 0.2, 0.7, 1.0], | ||
] | ||
) | ||
|
||
result = cytotrace.threshold_and_normalize_similarity_matrix(sim) | ||
|
||
# check diagonal is zeroed out | ||
assert np.all(np.diag(result) == 0) | ||
|
||
# check values below or equal to mean are zeroed out | ||
mean_sim = np.mean(sim) | ||
assert np.all(result[sim <= mean_sim] == 0) | ||
|
||
# check non-zero rows are normalized to sum to 1 | ||
non_zero_rows = np.where(result.sum(axis=1) > 0)[0] | ||
for row in non_zero_rows: | ||
assert_array_almost_equal(result[row].sum(), 1.0, decimal=6) | ||
|
||
# check zero rows remain zero | ||
zero_rows = np.where(result.sum(axis=1) == 0)[0] | ||
assert np.all(result[zero_rows] == 0) | ||
|
||
# check the result is sparse (contains zeros) | ||
assert np.sum(result == 0) > 0 | ||
|
||
# check the result preserves symmetry | ||
if np.allclose(sim, sim.T): | ||
assert np.allclose(result, result.T) | ||
|
||
# check stronger similarities are preserved | ||
stronger_similarities = sim > np.mean(sim) | ||
assert np.all(result[stronger_similarities] >= 0) | ||
|
||
# check weaker similarities are removed | ||
weaker_similarities = sim <= np.mean(sim) | ||
assert np.all(result[weaker_similarities] == 0) | ||
|
||
# check behavior with all-zero input | ||
zero_sim = np.zeros_like(sim) | ||
zero_result = cytotrace.threshold_and_normalize_similarity_matrix(zero_sim) | ||
assert np.all(zero_result == 0) | ||
|
||
# check behavior with negative values | ||
neg_sim = np.array([[-1, 0.5], [0.5, -1]]) | ||
neg_result = cytotrace.threshold_and_normalize_similarity_matrix(neg_sim) | ||
assert np.all(neg_result >= 0) | ||
assert_array_almost_equal(neg_result, np.array([[0, 1], [1, 0]])) | ||
|
||
|
||
def test_diffused(): | ||
markov = np.array([[0.7, 0.2, 0.1], [0.3, 0.5, 0.2], [0.1, 0.3, 0.6]]) | ||
gcs = np.array([1, 2, 3]) | ||
result = cytotrace.diffused(markov, gcs) | ||
assert result.shape == gcs.shape | ||
|
||
|
||
def test_cytotrace_sparse(small_anndata, monkeypatch): | ||
result = cytotrace.cytotrace_sparse(small_anndata, layer="raw") | ||
|
||
assert isinstance(result, dict) | ||
assert "CytoTRACE" in result | ||
assert "GCS" in result | ||
assert "cytoGenes" in result | ||
|
||
assert "gcs" in small_anndata.obs.columns | ||
assert "cytotrace" in small_anndata.obs.columns | ||
assert "counts" in small_anndata.obs.columns | ||
assert "cytotrace" in small_anndata.var.columns | ||
assert "cytotrace_corrs" in small_anndata.var.columns | ||
|
||
|
||
def test_cytotrace_sparse_errors(): | ||
adata = AnnData(X=np.array([[1, 2], [3, 4]])) | ||
adata.layers["raw"] = adata.X | ||
|
||
with pytest.raises( | ||
NotImplementedError, | ||
): | ||
cytotrace.cytotrace_sparse(adata) | ||
|
||
# print(cytotrace.__file__) | ||
with pytest.raises( | ||
KeyError, | ||
): | ||
cytotrace.cytotrace_sparse(adata, layer="non_existent") |