napistu 0.4.2__py3-none-any.whl → 0.4.4__py3-none-any.whl

tests/test_source.py

@@ -5,6 +5,8 @@ import os
 import pandas as pd
 from napistu import indices
 from napistu import source
+from napistu.network import ng_utils
+from napistu.constants import SBML_DFS
 
 test_path = os.path.abspath(os.path.join(__file__, os.pardir))
 test_data = os.path.join(test_path, "test_data")
@@ -58,10 +60,40 @@ def test_source_w_pwindex():
     assert source_obj.source.shape == (2, 8)
 
 
-################################################
-# __main__
-################################################
+def test_get_minimal_source_edges(sbml_dfs_metabolism):
+    vertices = sbml_dfs_metabolism.reactions.reset_index().rename(
+        columns={SBML_DFS.R_ID: "node"}
+    )
+
+    minimal_source_edges = ng_utils.get_minimal_sources_edges(
+        vertices, sbml_dfs_metabolism
+    )
+    # print(minimal_source_edges.shape)
+    assert minimal_source_edges.shape == (87, 3)
+
+
+def test_source_set_coverage(sbml_dfs_metabolism):
+
+    source_df = source.unnest_sources(sbml_dfs_metabolism.reactions)
+
+    # print(source_df.shape)
+    assert source_df.shape == (111, 7)
+
+    set_coverage = source.source_set_coverage(source_df)
+    # print(set_coverage.shape)
+    assert set_coverage.shape == (87, 6)
+
+
+def test_source_set_coverage_enrichment(sbml_dfs_metabolism):
+
+    source_total_counts = source.get_source_total_counts(
+        sbml_dfs_metabolism, "reactions"
+    )
+
+    source_df = source.unnest_sources(sbml_dfs_metabolism.reactions).head(40)
+
+    set_coverage = source.source_set_coverage(
+        source_df, source_total_counts=source_total_counts, sbml_dfs=sbml_dfs_metabolism
+    )
 
-if __name__ == "__main__":
-    test_source()
-    test_source_w_pwindex()
+    assert set_coverage.shape == (30, 6)

tests/test_statistics_hypothesis_testing.py

@@ -0,0 +1,62 @@
+import numpy as np
+from scipy.stats import fisher_exact
+
+from napistu.statistics import hypothesis_testing
+
+
+def test_fisher_exact_vectorized_basic_and_vectorized():
+
+    # Classic Fisher's test example: [[1, 9], [11, 3]]
+    # a=1, b=9, c=11, d=3
+    odds, p = hypothesis_testing.fisher_exact_vectorized([1], [9], [11], [3])
+    # Odds ratio: (1*3)/(9*11) = 3/99 = 0.0303...
+    assert np.allclose(odds, [3 / 99])
+    assert p.shape == (1,)
+    assert (p >= 0).all() and (p <= 1).all()
+
+    # Vectorized: two tables
+    odds, p = hypothesis_testing.fisher_exact_vectorized(
+        [1, 2], [9, 8], [11, 10], [3, 4]
+    )
+    assert odds.shape == (2,)
+    assert p.shape == (2,)
+    # Check that odds ratios are correct
+    expected_odds = np.array([(1 * 3) / (9 * 11), (2 * 4) / (8 * 10)])
+    assert np.allclose(odds, expected_odds)
+    # P-values should be between 0 and 1
+    assert (p >= 0).all() and (p <= 1).all()
+
+
+def test_fisher_exact_vectorized_vs_scipy():
+
+    # Define several 2x2 tables
+    tables = [
+        ([1], [9], [11], [3]),
+        ([5], [2], [8], [7]),
+        ([10], [10], [10], [10]),
+        ([0], [5], [5], [10]),
+        ([3], [7], [2], [8]),
+    ]
+    for a, b, c, d in tables:
+        odds_vec, p_vec = hypothesis_testing.fisher_exact_vectorized(a, b, c, d)
+        # Build the table for scipy
+        table = np.array([[a[0], b[0]], [c[0], d[0]]])
+        odds_scipy, p_scipy = fisher_exact(table, alternative="greater")
+        # Odds ratios should be nearly identical
+        assert np.allclose(odds_vec, [odds_scipy], rtol=1e-6, atol=1e-8)
+        # P-values should be close (normal approx vs exact)
+        assert np.allclose(
+            p_vec, [p_scipy], rtol=0.15, atol=1e-3
+        )  # allow some tolerance
+
+    # Also test vectorized input
+    a = [1, 5, 10, 0, 3]
+    b = [9, 2, 10, 5, 7]
+    c = [11, 8, 10, 5, 2]
+    d = [3, 7, 10, 10, 8]
+    odds_vec, p_vec = hypothesis_testing.fisher_exact_vectorized(a, b, c, d)
+    for i in range(len(a)):
+        table = np.array([[a[i], b[i]], [c[i], d[i]]])
+        odds_scipy, p_scipy = fisher_exact(table, alternative="greater")
+        assert np.allclose(odds_vec[i], odds_scipy, rtol=1e-6, atol=1e-8)
+        assert np.allclose(p_vec[i], p_scipy, rtol=0.15, atol=1e-3)

tests/test_set_coverage.py

@@ -1,50 +0,0 @@
-from __future__ import annotations
-
-from napistu import source
-from napistu.network import ng_utils
-
-
-def test_get_minimal_source_edges(sbml_dfs_metabolism):
-    vertices = sbml_dfs_metabolism.reactions.reset_index().rename(
-        columns={"r_id": "node"}
-    )
-
-    minimal_source_edges = ng_utils.get_minimal_sources_edges(
-        vertices, sbml_dfs_metabolism
-    )
-    # print(minimal_source_edges.shape)
-    assert minimal_source_edges.shape == (87, 3)
-
-
-def test_greedy_set_coverge_of_sources(sbml_dfs_metabolism):
-    table_schema = sbml_dfs_metabolism.schema["reactions"]
-
-    source_df = source.unnest_sources(
-        sbml_dfs_metabolism.reactions, source_var="r_Source"
-    )
-    # print(source_df.shape)
-    assert source_df.shape == (111, 7)
-
-    set_coverage = source.greedy_set_coverge_of_sources(source_df, table_schema)
-    # print(set_coverage.shape)
-    assert set_coverage.shape == (87, 6)
-
-
-################################################
-# __main__
-################################################
-
-if __name__ == "__main__":
-    import os
-    from napistu import indices
-    from napistu import consensus
-
-    test_path = os.path.abspath(os.path.join(__file__, os.pardir))
-    test_data = os.path.join(test_path, "test_data")
-
-    pw_index = indices.PWIndex(os.path.join(test_data, "pw_index_metabolism.tsv"))
-    sbml_dfs_dict = consensus.construct_sbml_dfs_dict(pw_index)
-    sbml_dfs_metabolism = consensus.construct_consensus_model(sbml_dfs_dict, pw_index)
-
-    test_get_minimal_source_edges(sbml_dfs_metabolism)
-    test_greedy_set_coverge_of_sources(sbml_dfs_metabolism)