Merge branch 'issue279' into issue275

nadeemlab · Dec 17, 2023 · 2b46c77 · 2b46c77
2 parents 588a3bd + 9720bdb
commit 2b46c77
Show file tree

Hide file tree

Showing 9 changed files with 17 additions and 16 deletions.
diff --git a/spatialprofilingtoolbox/cggnn/util.py b/spatialprofilingtoolbox/cggnn/util.py
@@ -15,7 +15,7 @@
 )
 from numpy.random import seed as np_seed
 from numpy.typing import NDArray
-from scipy.sparse import spmatrix
+from scipy.sparse import spmatrix, save_npz, load_npz  # type: ignore
 
 SETS = ('train', 'validation', 'test')
 SETS_type = Literal['train', 'validation', 'test']
@@ -47,19 +47,29 @@ class GraphData(NamedTuple):
 
 
 def save_hs_graphs(graphs_data: list[GraphData], output_directory: str) -> None:
-    """Save histological structure graphs to a directory."""
+    """Save histological structure graphs to a directory.
+
+    Saves the adjacency graph separately from the rest of the graph data for compatibility.
+    """
     makedirs(output_directory, exist_ok=True)
+    for i, gd in enumerate(graphs_data):
+        save_npz(join(output_directory, f'graph_{i}_adj.npz'), gd.graph.adj)
     with open(join(output_directory, 'graphs.pkl'), 'wb') as f:
+        for gd in graphs_data:
+            gd.graph.adj = None  # type: ignore
         dump(graphs_data, f)
 
 
 def load_hs_graphs(graph_directory: str) -> tuple[list[GraphData], list[str]]:
     """Load histological structure graphs from a directory.
 
-    Assumes directory contains the files `graphs.pkl` and `feature_names.txt`.
+    Assumes directory contains the files `graphs.pkl`, `feature_names.txt`, and a sparse array for
+    every graph in `graphs.pkl`.
     """
     with open(join(graph_directory, 'graphs.pkl'), 'rb') as f:
         graphs_data: list[GraphData] = load(f)
+    for i, gd in enumerate(graphs_data):
+        gd.graph.adj = load_npz(join(graph_directory, f'graph_{i}_adj.npz'))
     feature_names: list[str] = loadtxt(
         join(graph_directory, 'feature_names.txt'),
         dtype=str,

diff --git a/spatialprofilingtoolbox/workflow/assets/cggnn.nf b/spatialprofilingtoolbox/workflow/assets/cggnn.nf
@@ -116,8 +116,6 @@ process finalize_graphs {
 
     output:
     path "results/",                     emit: working_directory
-    path "results/feature_names.txt",    emit: feature_names_file
-    path "results/graphs.pkl",           emit: graphs_file
 
     script:
     """
@@ -153,8 +151,7 @@ process train {
 
     output:
     path "${working_directory}/importances.csv",     emit: importances_csv_path
-    path "${working_directory}/model/",              emit: model_directory
-    path "${working_directory}/graphs.pkl",          optional: true
+    path "${working_directory}/",                    emit: working_directory
 
     script:
     """
@@ -293,12 +290,6 @@ workflow {
     finalize_out.working_directory
         .set{ working_directory_ch }
 
-    finalize_out.feature_names_file
-        .set{ feature_names_ch }
-
-    finalize_out.graphs_file
-        .set{ graphs_file_ch }
-
     train(
         working_directory_ch,
         in_ram_ch,
@@ -315,8 +306,8 @@ workflow {
     train_out.importances_csv_path
         .set{ importances_csv_path_ch }
 
-    train_out.model_directory
-        .set{ model_directory_ch }
+    train_out.working_directory
+        .set{ working_directory_ch }
 
     upload_importance_scores(
         upload_importances_ch,

diff --git a/test/cggnn/unit_tests/feature_names.txt → ...cggnn/unit_tests/graphs/feature_names.txt b/test/cggnn/unit_tests/feature_names.txt → ...cggnn/unit_tests/graphs/feature_names.txt
diff --git a/test/cggnn/unit_tests/graphs/graph_0_adj.npz b/test/cggnn/unit_tests/graphs/graph_0_adj.npz
diff --git a/test/cggnn/unit_tests/graphs/graph_1_adj.npz b/test/cggnn/unit_tests/graphs/graph_1_adj.npz
diff --git a/test/cggnn/unit_tests/graphs/graph_2_adj.npz b/test/cggnn/unit_tests/graphs/graph_2_adj.npz
diff --git a/test/cggnn/unit_tests/graphs/graph_3_adj.npz b/test/cggnn/unit_tests/graphs/graph_3_adj.npz
diff --git a/test/cggnn/unit_tests/graphs.pkl → test/cggnn/unit_tests/graphs/graphs.pkl b/test/cggnn/unit_tests/graphs.pkl → test/cggnn/unit_tests/graphs/graphs.pkl
diff --git a/test/cggnn/unit_tests/test_plotting.sh b/test/cggnn/unit_tests/test_plotting.sh
@@ -1,5 +1,5 @@
 spt cggnn plot-interactives \
-    --cg_path unit_tests/ \
+    --cg_path unit_tests/graphs/ \
     --output_directory . \
     --merge_rois
 plotting_ran="$?"