fix failing clustering tests

mdpath/src/cluster.py

@@ -117,7 +117,7 @@ def pathways_cluster(
         linkage_matrix = hierarchy.linkage(distance_matrix.values, method="complete")
 
         silhouette_scores = []
-        for n_clusters in range(2, len(overlap_matrix) + 1):
+        for n_clusters in range(2, len(overlap_matrix)):
             cluster_labels = hierarchy.fcluster(
                 linkage_matrix, n_clusters, criterion="maxclust"
             )

mdpath/tests/test_cluster.py

@@ -70,24 +70,24 @@ def setup_clustering(load_test_data):
 
     return PatwayClustering(df_close_res, top_pathways, num_parallel_processes)
 
-#def test_pathways_cluster(setup_clustering):
-#    clustering = setup_clustering
-#
-#    with patch("matplotlib.pyplot.savefig") as mock_savefig:
-#        clusters = clustering.pathways_cluster(n_top_clust=3)#
-#
-#    assert isinstance(clusters, dict)
-#    assert all(isinstance(k, int) and isinstance(v, list) for k, v in clusters.items())
-#    assert len(clusters) > 0
-#
-#    current_dir = os.path.dirname(os.path.abspath(__file__))
-#    file_path = os.path.join(current_dir, "clusters.pkl")
-#    with open(file_path, "rb") as file:
-#        saved_clusters = pickle.load(file)
-#
-#    assert clusters == saved_clusters
-#
-#    mock_savefig.assert_called_once_with("clustered_paths.png")
+def test_pathways_cluster(setup_clustering):
+   clustering = setup_clustering
+
+   with patch("matplotlib.pyplot.savefig") as mock_savefig:
+       clusters = clustering.pathways_cluster(n_top_clust=3)#
+
+   assert isinstance(clusters, dict)
+   assert all(isinstance(k, int) and isinstance(v, list) for k, v in clusters.items())
+   assert len(clusters) > 0
+
+   current_dir = os.path.dirname(os.path.abspath(__file__))
+   file_path = os.path.join(current_dir, "clusters.pkl")
+   with open(file_path, "rb") as file:
+       saved_clusters = pickle.load(file)
+
+   assert clusters == saved_clusters
+
+   mock_savefig.assert_called_once_with("clustered_paths.png")
 
 
 @pytest.fixture(scope="module")

mdpath/tests/test_mdpath.py

@@ -80,62 +80,62 @@ def test_mdpath_wrong_input(tmp_path):
         os.chdir(original_cwd)
 
 
-#def test_mdpath_output_files():
-#    script_dir = os.path.dirname(os.path.abspath(__file__))
-#    project_root = os.path.dirname(script_dir)
-#    mdpath_dir = os.path.join(project_root, "mdpath")
-#
-#    topology = os.path.join(script_dir, "test_topology.pdb")
-#    trajectory = os.path.join(script_dir, "test_trajectory.dcd")
-#    numpath = "10" 
-#    bootstrap = "1"
-#    assert os.path.exists(topology), f"Topology file {topology} does not exist."
-#    assert os.path.exists(trajectory), f"Trajectory file {trajectory} does not exist."
-#
-#    expected_files = [
-#        os.path.join(script_dir, "first_frame.pdb"),
-#        os.path.join(script_dir, "nmi_df.csv"),
-#        os.path.join(script_dir, "output.txt"),
-#        os.path.join(script_dir, "residue_coordinates.pkl"),
-#        os.path.join(script_dir, "cluster_pathways_dict.pkl"),
-#        os.path.join(script_dir, "clusters_paths.json"),
-#        os.path.join(script_dir, "precomputed_clusters_paths.json"),
-#        os.path.join(script_dir, "quick_precomputed_clusters_paths.json"),
-#        os.path.join(script_dir, "bootstrap/bootstrap_sample_0.txt"),
-#    ]
-#
-#    sys.path.insert(0, mdpath_dir)
-#
-#    try:
-#        from mdpath.mdpath import main as mdpath_main
-#    except ImportError as e:
-#        raise ImportError(f"Error importing mdpath: {e}")
-#
-#    original_cwd = os.getcwd()
-#    os.chdir(script_dir)
-#
-#    try:
-#        sys.argv = [
-#            "mdpath",
-#            "-top",
-#            topology,
-#            "-traj",
-#            trajectory,
-#            "-numpath",
-#            numpath,
-#            "-bs",
-#            bootstrap,
-#            "-lig",
-#            "272",
-#        ]
-#
-#        mdpath_main()
-#
-#        for file in expected_files:
-#            assert os.path.exists(file), f"Expected output file {file} not found."
-#
-#    finally:
-#        for file in expected_files:
-#            if os.path.exists(file):
-#                os.remove(file)
-#        os.chdir(original_cwd)
+def test_mdpath_output_files():
+   script_dir = os.path.dirname(os.path.abspath(__file__))
+   project_root = os.path.dirname(script_dir)
+   mdpath_dir = os.path.join(project_root, "mdpath")
+
+   topology = os.path.join(script_dir, "test_topology.pdb")
+   trajectory = os.path.join(script_dir, "test_trajectory.dcd")
+   numpath = "10" 
+   bootstrap = "1"
+   assert os.path.exists(topology), f"Topology file {topology} does not exist."
+   assert os.path.exists(trajectory), f"Trajectory file {trajectory} does not exist."
+
+   expected_files = [
+       os.path.join(script_dir, "first_frame.pdb"),
+       os.path.join(script_dir, "nmi_df.csv"),
+       os.path.join(script_dir, "output.txt"),
+       os.path.join(script_dir, "residue_coordinates.pkl"),
+       os.path.join(script_dir, "cluster_pathways_dict.pkl"),
+       os.path.join(script_dir, "clusters_paths.json"),
+       os.path.join(script_dir, "precomputed_clusters_paths.json"),
+       os.path.join(script_dir, "quick_precomputed_clusters_paths.json"),
+       os.path.join(script_dir, "bootstrap/bootstrap_sample_0.txt"),
+   ]
+
+   sys.path.insert(0, mdpath_dir)
+
+   try:
+       from mdpath.mdpath import main as mdpath_main
+   except ImportError as e:
+       raise ImportError(f"Error importing mdpath: {e}")
+
+   original_cwd = os.getcwd()
+   os.chdir(script_dir)
+
+   try:
+       sys.argv = [
+           "mdpath",
+           "-top",
+           topology,
+           "-traj",
+           trajectory,
+           "-numpath",
+           numpath,
+           "-bs",
+           bootstrap,
+           "-lig",
+           "272",
+       ]
+
+       mdpath_main()
+
+       for file in expected_files:
+           assert os.path.exists(file), f"Expected output file {file} not found."
+
+   finally:
+       for file in expected_files:
+           if os.path.exists(file):
+               os.remove(file)
+       os.chdir(original_cwd)

mdpath/tests/test_mdpath_tools.py

@@ -265,40 +265,40 @@ def test_path_comparison():
             os.remove(file_path)
 
 
-#def test_multitraj_analysis():
-#    script_dir = os.path.dirname(os.path.abspath(__file__))
-#    current_directory = os.getcwd()
-#    topology = os.path.join(script_dir, "multitraj.pdb")
-#    multitraj_1 = os.path.join(script_dir, "top_pathways.pkl")
-#
-#    original_stdout = sys.stdout
-#    sys.stdout = StringIO()
-#
-#    generated_files = []  # Define an empty list for generated_files
-#
-#    try:
-#        sys.argv = [
-#            "mdpath_multitraj",
-#            "-top",
-#            topology,
-#            "-multitraj",
-#            multitraj_1,
-#            multitraj_1
-#        ]
-#        with pytest.raises(SystemExit) as exc_info:
-#            multitraj_analysis()
-#
-#        assert exc_info.value.code == 0, "The command failed with non-zero exit code"
-#
-#
-#        generated_files = glob.glob(
-#            os.path.join(current_directory, "multitraj_clusters_paths.json")
-#        )
-#        assert (
-#            len(generated_files) > 0
-#        ), "No multitraj_clusters_paths.json file was generated."
-#
-#    finally:
-#        sys.stdout = original_stdout
-#        for file_path in generated_files:
-#            os.remove(file_path)
+def test_multitraj_analysis():
+   script_dir = os.path.dirname(os.path.abspath(__file__))
+   current_directory = os.getcwd()
+   topology = os.path.join(script_dir, "multitraj.pdb")
+   multitraj_1 = os.path.join(script_dir, "top_pathways.pkl")
+
+   original_stdout = sys.stdout
+   sys.stdout = StringIO()
+
+   generated_files = []  # Define an empty list for generated_files
+
+   try:
+       sys.argv = [
+           "mdpath_multitraj",
+           "-top",
+           topology,
+           "-multitraj",
+           multitraj_1,
+           multitraj_1
+       ]
+       with pytest.raises(SystemExit) as exc_info:
+           multitraj_analysis()
+
+       assert exc_info.value.code == 0, "The command failed with non-zero exit code"
+
+
+       generated_files = glob.glob(
+           os.path.join(current_directory, "multitraj_clusters_paths.json")
+       )
+       assert (
+           len(generated_files) > 0
+       ), "No multitraj_clusters_paths.json file was generated."
+
+   finally:
+       sys.stdout = original_stdout
+       for file_path in generated_files:
+           os.remove(file_path)