test(cytotrace): add test module

Signed-off-by: Cameron Smith <[email protected]>

src/pyrovelocity/tests/analysis/test_cytotrace.py

@@ -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")