napistu 0.3.6__py3-none-any.whl → 0.4.0__py3-none-any.whl

tests/test_network_net_create_utils.py

@@ -0,0 +1,538 @@
+from __future__ import annotations
+
+import pytest
+
+import pandas as pd
+
+from napistu.network import net_create_utils
+from napistu.constants import (
+    MINI_SBO_FROM_NAME,
+    SBML_DFS,
+    SBOTERM_NAMES,
+    VALID_SBO_TERM_NAMES,
+)
+from napistu.network.constants import (
+    DROP_REACTIONS_WHEN,
+    NAPISTU_GRAPH_EDGES,
+    NAPISTU_GRAPH_NODE_TYPES,
+    VALID_GRAPH_WIRING_APPROACHES,
+)
+
+
+def test_format_interactors(reaction_species_examples):
+
+    r_id = "foo"
+    # interactions are formatted
+    graph_hierarchy_df = net_create_utils.create_graph_hierarchy_df("regulatory")
+
+    assert (
+        net_create_utils.format_tiered_reaction_species(
+            reaction_species_examples["valid_interactor"],
+            r_id,
+            graph_hierarchy_df,
+            drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER,
+        ).shape[0]
+        == 1
+    )
+
+    # simple reaction with just substrates and products
+    assert (
+        net_create_utils.format_tiered_reaction_species(
+            reaction_species_examples["sub_and_prod"],
+            r_id,
+            graph_hierarchy_df,
+            drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER,
+        ).shape[0]
+        == 2
+    )
+
+    # add a stimulator (activator)
+    rxn_edges = net_create_utils.format_tiered_reaction_species(
+        reaction_species_examples["stimulator"],
+        r_id,
+        graph_hierarchy_df,
+        drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER,
+    )
+
+    assert rxn_edges.shape[0] == 3
+    assert rxn_edges.iloc[0][["from", "to"]].tolist() == ["stim", "sub"]
+
+    # add catalyst + stimulator
+    rxn_edges = net_create_utils.format_tiered_reaction_species(
+        reaction_species_examples["all_entities"],
+        r_id,
+        graph_hierarchy_df,
+        drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER,
+    )
+
+    assert rxn_edges.shape[0] == 4
+    assert rxn_edges.iloc[0][["from", "to"]].tolist() == ["stim", "cat"]
+    assert rxn_edges.iloc[1][["from", "to"]].tolist() == ["cat", "sub"]
+
+    # no substrate
+    rxn_edges = net_create_utils.format_tiered_reaction_species(
+        reaction_species_examples["no_substrate"],
+        r_id,
+        graph_hierarchy_df,
+        drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER,
+    )
+
+    assert rxn_edges.shape[0] == 5
+    # stimulator -> reactant
+    assert rxn_edges.iloc[0][["from", "to"]].tolist() == ["stim1", "cat"]
+    assert rxn_edges.iloc[1][["from", "to"]].tolist() == ["stim2", "cat"]
+    assert rxn_edges.iloc[2][["from", "to"]].tolist() == ["inh", "cat"]
+
+    # use the surrogate model tiered layout also
+
+    graph_hierarchy_df = net_create_utils.create_graph_hierarchy_df("surrogate")
+
+    rxn_edges = net_create_utils.format_tiered_reaction_species(
+        reaction_species_examples["all_entities"],
+        r_id,
+        graph_hierarchy_df,
+        drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER,
+    )
+
+    assert rxn_edges.shape[0] == 4
+    assert rxn_edges.iloc[0][["from", "to"]].tolist() == ["stim", "sub"]
+    assert rxn_edges.iloc[1][["from", "to"]].tolist() == ["sub", "cat"]
+
+
+def test_drop_reactions_when_parameters(reaction_species_examples):
+    """Test different drop_reactions_when parameter values and edge cases."""
+
+    r_id = "foo"
+    graph_hierarchy = net_create_utils.create_graph_hierarchy_df("regulatory")
+
+    # Test ALWAYS - should drop reaction regardless of tiers
+    edges_always = net_create_utils.format_tiered_reaction_species(
+        reaction_species_examples["all_entities"],
+        r_id,
+        graph_hierarchy,
+        DROP_REACTIONS_WHEN.ALWAYS,
+    )
+    assert r_id not in edges_always[NAPISTU_GRAPH_EDGES.FROM].values
+    assert r_id not in edges_always[NAPISTU_GRAPH_EDGES.TO].values
+
+    # Test EDGELIST with 2 species (should drop reaction)
+    edges_edgelist = net_create_utils.format_tiered_reaction_species(
+        reaction_species_examples["sub_and_prod"],
+        r_id,
+        graph_hierarchy,
+        DROP_REACTIONS_WHEN.EDGELIST,
+    )
+    assert r_id not in edges_edgelist[NAPISTU_GRAPH_EDGES.FROM].values
+
+    # Test EDGELIST with >2 species (should keep reaction)
+    edges_multi = net_create_utils.format_tiered_reaction_species(
+        reaction_species_examples["all_entities"],
+        r_id,
+        graph_hierarchy,
+        DROP_REACTIONS_WHEN.EDGELIST,
+    )
+    reaction_in_edges = (
+        r_id in edges_multi[NAPISTU_GRAPH_EDGES.FROM].values
+        or r_id in edges_multi[NAPISTU_GRAPH_EDGES.TO].values
+    )
+    assert reaction_in_edges
+
+    # Test invalid parameter
+    with pytest.raises(ValueError, match="Invalid drop_reactions"):
+        net_create_utils.format_tiered_reaction_species(
+            reaction_species_examples["sub_and_prod"],
+            r_id,
+            graph_hierarchy,
+            "INVALID_OPTION",
+        )
+
+
+def test_edge_cases_and_validation(reaction_species_examples):
+    """Test edge cases, empty inputs, and validation errors."""
+
+    r_id = "foo"
+    graph_hierarchy = net_create_utils.create_graph_hierarchy_df("regulatory")
+
+    # Test single species
+    edges_single = net_create_utils.format_tiered_reaction_species(
+        reaction_species_examples["single_species"], r_id, graph_hierarchy
+    )
+    assert edges_single.empty
+
+    # Test validation with incorrectly indexed DataFrame (should raise error)
+    bad_df = reaction_species_examples[
+        "sub_and_prod"
+    ].reset_index()  # Remove proper index
+    with pytest.raises(ValueError):
+        net_create_utils.format_tiered_reaction_species(bad_df, r_id, graph_hierarchy)
+
+    # Test activator and inhibitor only (should return empty DataFrame)
+    edges_ai = net_create_utils.format_tiered_reaction_species(
+        reaction_species_examples["activator_and_inhibitor_only"], r_id, graph_hierarchy
+    )
+    assert edges_ai.empty
+
+
+def test_edgelist_should_have_one_edge():
+    """EDGELIST with 2 species should create exactly 1 edge, not 2"""
+
+    r_id = "foo"
+    reaction_df = pd.DataFrame(
+        {
+            SBML_DFS.SBO_TERM: [
+                MINI_SBO_FROM_NAME[SBOTERM_NAMES.REACTANT],
+                MINI_SBO_FROM_NAME[SBOTERM_NAMES.PRODUCT],
+            ],
+            SBML_DFS.SC_ID: ["sub", "prod"],
+            SBML_DFS.STOICHIOMETRY: [-1.0, 1.0],
+        }
+    ).set_index(SBML_DFS.SBO_TERM)
+
+    graph_hierarchy = net_create_utils.create_graph_hierarchy_df("regulatory")
+    edges = net_create_utils.format_tiered_reaction_species(
+        reaction_df, r_id, graph_hierarchy, DROP_REACTIONS_WHEN.EDGELIST
+    )
+
+    # Should be 1 edge, actually gets 2
+    assert len(edges) == 1, f"EDGELIST should create 1 edge, got {len(edges)}"
+
+
+def test_edgelist_should_not_have_reaction_as_source():
+    """EDGELIST should not have reaction ID in FROM column"""
+
+    r_id = "foo"
+    reaction_df = pd.DataFrame(
+        {
+            SBML_DFS.SBO_TERM: [
+                MINI_SBO_FROM_NAME[SBOTERM_NAMES.REACTANT],
+                MINI_SBO_FROM_NAME[SBOTERM_NAMES.PRODUCT],
+            ],
+            SBML_DFS.SC_ID: ["sub", "prod"],
+            SBML_DFS.STOICHIOMETRY: [-1.0, 1.0],
+        }
+    ).set_index(SBML_DFS.SBO_TERM)
+
+    graph_hierarchy = net_create_utils.create_graph_hierarchy_df("regulatory")
+    edges = net_create_utils.format_tiered_reaction_species(
+        reaction_df, r_id, graph_hierarchy, DROP_REACTIONS_WHEN.EDGELIST
+    )
+
+    # Should not have 'foo' in FROM column, but it does
+    assert (
+        r_id not in edges["from"].values
+    ), f"Reaction {r_id} should not appear in FROM column"
+
+
+def test_should_drop_reaction(reaction_species_examples):
+
+    r_id = "foo"
+
+    graph_hierarchy_df = net_create_utils.create_graph_hierarchy_df("regulatory")
+
+    rxn_species = reaction_species_examples["sub_and_prod"]
+    net_create_utils._validate_sbo_indexed_rsc_stoi(rxn_species)
+
+    # map reaction species to the tiers of the graph hierarchy. higher levels point to lower levels
+    # same-level entries point at each other only if there is only a single tier
+    entities_ordered_by_tier = net_create_utils._reaction_species_to_tiers(
+        rxn_species, graph_hierarchy_df, r_id
+    )
+
+    # this is an edgeliist (just 2 entries)
+    assert net_create_utils._should_drop_reaction(
+        entities_ordered_by_tier, drop_reactions_when=DROP_REACTIONS_WHEN.EDGELIST
+    )
+
+    # not the same tier
+    assert not net_create_utils._should_drop_reaction(
+        entities_ordered_by_tier, drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER
+    )
+
+
+def test_graph_hierarchy_layouts():
+    REQUIRED_NAMES = VALID_SBO_TERM_NAMES + [NAPISTU_GRAPH_NODE_TYPES.REACTION]
+    for value in VALID_GRAPH_WIRING_APPROACHES:
+        layout_df = net_create_utils.create_graph_hierarchy_df(value)
+        # all terms should be represented
+        missing = set(REQUIRED_NAMES).difference(
+            set(layout_df[NAPISTU_GRAPH_EDGES.SBO_NAME])
+        )
+        assert not missing, f"Missing SBO names in {value}: {missing}"
+        # all terms should be unique
+        duplicated = layout_df[layout_df[NAPISTU_GRAPH_EDGES.SBO_NAME].duplicated()]
+        assert (
+            duplicated.empty
+        ), f"Duplicated SBO names in {value}: {duplicated[NAPISTU_GRAPH_EDGES.SBO_NAME].tolist()}"
+        # check that reaction is present and its by itself
+        reaction_tiers = layout_df[
+            layout_df[NAPISTU_GRAPH_EDGES.SBO_NAME] == NAPISTU_GRAPH_NODE_TYPES.REACTION
+        ]["tier"].unique()
+        assert (
+            len(reaction_tiers) == 1
+        ), f"'reaction' appears in multiple tiers in {value}: {reaction_tiers}"
+        reaction_tier = reaction_tiers[0]
+        reaction_tier_df = layout_df[layout_df["tier"] == reaction_tier]
+        assert (
+            reaction_tier_df.shape[0] == 1
+            and reaction_tier_df[NAPISTU_GRAPH_EDGES.SBO_NAME].iloc[0]
+            == NAPISTU_GRAPH_NODE_TYPES.REACTION
+        ), f"Tier {reaction_tier} in {value} should contain only 'reaction', but contains: {reaction_tier_df[NAPISTU_GRAPH_EDGES.SBO_NAME].tolist()}"
+
+
+def test_identifying_and_formatting_interactor_duos(reaction_species_examples):
+
+    # directly specify interactions as a speed up to same tier procedure
+    interaction_template = reaction_species_examples["valid_interactor"]
+    interactor_species = pd.concat(
+        [
+            interaction_template.reset_index().assign(
+                r_id=str(i),
+                sc_id=lambda df: df[SBML_DFS.SC_ID].apply(lambda x: f"r{i}_{x}"),
+            )
+            for i in range(0, 10)
+        ]
+    )
+
+    invalid_interactor_template = reaction_species_examples["invalid_interactor"]
+    invalid_interactor_species = pd.concat(
+        [
+            invalid_interactor_template.reset_index().assign(
+                r_id=str(i),
+                sc_id=lambda df: df[SBML_DFS.SC_ID].apply(lambda x: f"r{i}_{x}"),
+            )
+            for i in range(10, 12)
+        ]
+    )
+
+    reaction_species = pd.concat([interactor_species, invalid_interactor_species])
+
+    matching_r_ids = net_create_utils._find_sbo_duos(
+        reaction_species, MINI_SBO_FROM_NAME[SBOTERM_NAMES.INTERACTOR]
+    )
+    assert set(matching_r_ids) == {str(i) for i in range(0, 10)}
+
+    interactor_duos = reaction_species.loc[
+        reaction_species[SBML_DFS.R_ID].isin(matching_r_ids)
+    ]
+    assert net_create_utils._interactor_duos_to_wide(interactor_duos).shape[0] == 10
+
+
+def test_wire_reaction_species_mixed_interactions(reaction_species_examples):
+    """
+    Test wire_reaction_species function with a mix of interactor and non-interactor reactions.
+
+    This test verifies that the function correctly processes:
+    1. Interactor pairs (processed en-masse)
+    2. Non-interactor reactions (processed with tiered algorithms)
+    3. Different wiring approaches
+    4. Different drop_reactions_when conditions
+    """
+
+    # Create a mixed dataset with both interactor and non-interactor reactions
+    # Interactor reactions (should be processed en-masse)
+    interactor_template = reaction_species_examples["valid_interactor"]
+    interactor_species = pd.concat(
+        [
+            interactor_template.reset_index().assign(
+                r_id=f"interactor_{i}",
+                sc_id=lambda df: df[SBML_DFS.SC_ID].apply(
+                    lambda x: f"interactor_{i}_{x}"
+                ),
+            )
+            for i in range(3)  # 3 interactor reactions
+        ]
+    )
+
+    # Non-interactor reactions (should be processed with tiered algorithms)
+    non_interactor_species = pd.concat(
+        [
+            reaction_species_examples["sub_and_prod"]
+            .reset_index()
+            .assign(
+                r_id=f"reaction_{i}",
+                sc_id=lambda df: df[SBML_DFS.SC_ID].apply(
+                    lambda x: f"reaction_{i}_{x}"
+                ),
+            )
+            for i in range(2)  # 2 substrate-product reactions
+        ]
+    )
+
+    # Add a complex reaction with multiple entity types
+    complex_reaction = (
+        reaction_species_examples["all_entities"]
+        .reset_index()
+        .assign(
+            r_id="complex_reaction",
+            sc_id=lambda df: df[SBML_DFS.SC_ID].apply(lambda x: f"complex_{x}"),
+        )
+    )
+
+    # Combine all reaction species
+    all_reaction_species = pd.concat(
+        [interactor_species, non_interactor_species, complex_reaction]
+    )
+
+    # Test with regulatory wiring approach
+    edges_regulatory = net_create_utils.wire_reaction_species(
+        all_reaction_species,
+        wiring_approach="regulatory",
+        drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER,
+    )
+
+    # Verify the output structure
+    assert not edges_regulatory.empty
+    required_columns = [
+        NAPISTU_GRAPH_EDGES.FROM,
+        NAPISTU_GRAPH_EDGES.TO,
+        NAPISTU_GRAPH_EDGES.STOICHIOMETRY,
+        NAPISTU_GRAPH_EDGES.SBO_TERM,
+        SBML_DFS.R_ID,
+    ]
+
+    for col in required_columns:
+        assert col in edges_regulatory.columns, f"Missing required column: {col}"
+
+    # Check that interactor reactions were processed correctly
+    interactor_edges = edges_regulatory[
+        edges_regulatory[SBML_DFS.R_ID].str.startswith("interactor_")
+    ]
+    assert (
+        len(interactor_edges) == 3
+    ), f"Expected 3 interactor edges, got {len(interactor_edges)}"
+
+    # Check that non-interactor reactions were processed correctly
+    reaction_edges = edges_regulatory[
+        edges_regulatory[SBML_DFS.R_ID].str.startswith("reaction_")
+    ]
+    assert (
+        len(reaction_edges) == 4
+    ), f"Expected 4 reaction edges, got {len(reaction_edges)}"
+
+    # Check that complex reaction was processed correctly
+    complex_edges = edges_regulatory[
+        edges_regulatory[SBML_DFS.R_ID] == "complex_reaction"
+    ]
+    assert (
+        len(complex_edges) == 4
+    ), f"Expected 4 complex reaction edges, got {len(complex_edges)}"
+
+    # Test with different drop_reactions_when condition
+    edges_always_drop = net_create_utils.wire_reaction_species(
+        all_reaction_species,
+        wiring_approach="regulatory",
+        drop_reactions_when=DROP_REACTIONS_WHEN.ALWAYS,
+    )
+
+    # With ALWAYS drop, reaction IDs should not appear in FROM or TO columns
+    reaction_ids = [
+        "interactor_0",
+        "interactor_1",
+        "interactor_2",
+        "reaction_0",
+        "reaction_1",
+        "complex_reaction",
+    ]
+    for r_id in reaction_ids:
+        assert (
+            r_id not in edges_always_drop[NAPISTU_GRAPH_EDGES.FROM].values
+        ), f"Reaction {r_id} should not be in FROM column"
+        assert (
+            r_id not in edges_always_drop[NAPISTU_GRAPH_EDGES.TO].values
+        ), f"Reaction {r_id} should not be in TO column"
+
+    # Test with EDGELIST drop condition
+    edges_edgelist = net_create_utils.wire_reaction_species(
+        all_reaction_species,
+        wiring_approach="regulatory",
+        drop_reactions_when=DROP_REACTIONS_WHEN.EDGELIST,
+    )
+
+    # Simple reactions (2 species) should not have reaction IDs, but complex reactions should
+    simple_reaction_ids = ["reaction_0", "reaction_1"]
+    complex_reaction_id = "complex_reaction"
+
+    for r_id in simple_reaction_ids:
+        assert (
+            r_id not in edges_edgelist[NAPISTU_GRAPH_EDGES.FROM].values
+        ), f"Simple reaction {r_id} should not be in FROM column"
+        assert (
+            r_id not in edges_edgelist[NAPISTU_GRAPH_EDGES.TO].values
+        ), f"Simple reaction {r_id} should not be in TO column"
+
+    # Complex reaction should still have reaction ID in edges
+    complex_in_edges = (
+        complex_reaction_id in edges_edgelist[NAPISTU_GRAPH_EDGES.FROM].values
+        or complex_reaction_id in edges_edgelist[NAPISTU_GRAPH_EDGES.TO].values
+    )
+    assert (
+        complex_in_edges
+    ), f"Complex reaction {complex_reaction_id} should appear in edges"
+
+    # Test edge case: only interactor reactions
+    only_interactors = interactor_species
+    edges_only_interactors = net_create_utils.wire_reaction_species(
+        only_interactors,
+        wiring_approach="regulatory",
+        drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER,
+    )
+
+    assert not edges_only_interactors.empty
+    assert (
+        len(edges_only_interactors) == 3
+    ), f"Expected 3 edges for only interactors, got {len(edges_only_interactors)}"
+
+    # Test edge case: only non-interactor reactions
+    only_reactions = pd.concat([non_interactor_species, complex_reaction])
+    edges_only_reactions = net_create_utils.wire_reaction_species(
+        only_reactions,
+        wiring_approach="regulatory",
+        drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER,
+    )
+
+    print(edges_only_reactions)
+
+    assert not edges_only_reactions.empty
+    assert (
+        len(edges_only_reactions) == 8
+    ), f"Expected 8 edges for only reactions, got {len(edges_only_reactions)}"
+
+
+def test_wire_reaction_species_validation_errors():
+    """Test wire_reaction_species function with invalid inputs."""
+
+    # Test with invalid wiring approach
+    reaction_species = pd.DataFrame(
+        {
+            SBML_DFS.R_ID: ["R1"],
+            SBML_DFS.SC_ID: ["A"],
+            SBML_DFS.STOICHIOMETRY: [0],
+            SBML_DFS.SBO_TERM: ["SBO:0000336"],  # interactor
+        }
+    )
+
+    with pytest.raises(ValueError, match="is not a valid wiring approach"):
+        net_create_utils.wire_reaction_species(
+            reaction_species,
+            wiring_approach="invalid_approach",
+            drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER,
+        )
+
+    # Test with invalid SBO terms
+    invalid_sbo_species = pd.DataFrame(
+        {
+            SBML_DFS.R_ID: ["R1"],
+            SBML_DFS.SC_ID: ["A"],
+            SBML_DFS.STOICHIOMETRY: [0],
+            SBML_DFS.SBO_TERM: ["INVALID_SBO_TERM"],
+        }
+    )
+
+    with pytest.raises(
+        ValueError, match="Some reaction species have unusable SBO terms"
+    ):
+        net_create_utils.wire_reaction_species(
+            invalid_sbo_species,
+            wiring_approach="regulatory",
+            drop_reactions_when=DROP_REACTIONS_WHEN.SAME_TIER,
+        )

tests/test_network_ng_core.py

@@ -0,0 +1,19 @@
+import igraph as ig
+import logging
+
+from napistu.network.ng_core import NapistuGraph
+
+logger = logging.getLogger(__name__)
+
+
+def test_remove_isolated_vertices():
+    """Test removing isolated vertices from a graph."""
+
+    g = ig.Graph()
+    g.add_vertices(5, attributes={"name": ["A", "B", "C", "D", "E"]})
+    g.add_edges([(0, 1), (2, 3)])  # A-B, C-D connected; E isolated
+
+    napstu_graph = NapistuGraph.from_igraph(g)
+    napstu_graph.remove_isolated_vertices()
+    assert napstu_graph.vcount() == 4  # Should have 4 vertices after removing E
+    assert "E" not in [v["name"] for v in napstu_graph.vs]  # E should be gone

tests/test_network_ng_utils.py

@@ -2,7 +2,7 @@ import igraph as ig
 import pandas as pd
 
 from napistu.network import ng_utils
-from napistu.network.napistu_graph_core import NapistuGraph
+from napistu.network.ng_core import NapistuGraph
 
 
 def test_napistu_graph_to_pandas_dfs():

tests/test_network_precompute.py

@@ -7,6 +7,7 @@ import tempfile
 import numpy as np
 import pandas as pd
 from napistu import sbml_dfs_core
+from napistu import utils
 from napistu.ingestion import sbml
 from napistu.network import neighborhoods
 from napistu.network import net_create
@@ -18,12 +19,12 @@ sbml_path = os.path.join(test_path, "test_data", "reactome_glucose_metabolism.sb
 if not os.path.isfile(sbml_path):
     raise ValueError(f"{sbml_path} not found")
 
-sbml_model = sbml.SBML(sbml_path).model
+sbml_model = sbml.SBML(sbml_path)
 sbml_dfs = sbml_dfs_core.SBML_dfs(sbml_model)
 sbml_dfs.validate()
 
 napistu_graph = net_create.process_napistu_graph(
-    sbml_dfs, graph_type="bipartite", directed=True, weighting_strategy="topology"
+    sbml_dfs, wiring_approach="bipartite", directed=True, weighting_strategy="topology"
 )
 
 # number of species to include when finding all x all paths
@@ -263,10 +264,10 @@ def test_precomputed_distances_serialization():
 
     try:
         # Test serialization
-        precompute.save_precomputed_distances(original_df, temp_path)
+        utils.save_parquet(original_df, temp_path)
 
         # Test deserialization
-        loaded_df = precompute.load_precomputed_distances(temp_path)
+        loaded_df = utils.load_parquet(temp_path)
 
         # Validate that the loaded DataFrame is identical to the original
         pd.testing.assert_frame_equal(original_df, loaded_df, check_like=True)