Bug merge detection (#167)

* bug: overcount swcs

* refactor: updated merge/split rates

---------

Co-authored-by: anna-grim <anna.grim@alleninstitute.org>

Author: anna-grim

pyproject.toml

@@ -20,7 +20,7 @@ dependencies = [
     "networkx",
     "numpy",
     "pandas",
-    "s3fs==2025.6.0",
+    "s3fs",
     "scikit-image",
     "tensorstore",
     "tifffile",

src/segmentation_skeleton_metrics/skeleton_metrics.py

@@ -713,7 +713,8 @@ def __call__(self, gt_graphs, results):
         for name, graph in gt_graphs.items():
             # Compute result
             if results["# Splits"][name] > 0:
-                rate = graph.labeled_run_length / results["# Splits"][name]
+                rl = util.compute_segmented_run_length(graph, results)
+                rate = rl / results["# Splits"][name]
             else:
                 rate = np.nan
             new_results[name] = round(rate, 2)
@@ -766,7 +767,8 @@ def __call__(self, gt_graphs, results):
         for name, graph in gt_graphs.items():
             # Compute result
             if results["# Merges"][name] > 0:
-                rate = graph.labeled_run_length / results["# Merges"][name]
+                rl = util.compute_segmented_run_length(graph, results)
+                rate = rl / results["# Merges"][name]
             else:
                 rate = np.nan
             new_results[name] = round(rate, 2)

src/segmentation_skeleton_metrics/utils/util.py

@@ -483,6 +483,27 @@ def compute_weighted_avg(df, column_name):
         return (values * weights).sum() / weights.sum()
 
 
+def compute_segmented_run_length(graph, results):
+    """
+    Computes the run length of a graph that was segmented.
+
+    Parameters
+    ----------
+    graph : LabeledGraph
+        Graph to be evaluated.
+    results : pandas.DataFrame
+        Data frame containing skeleton metrics
+
+    Returns
+    -------
+    float
+        Run length of a graph that was segmented.
+    """
+    omit_run_length = graph.run_length * results["% Omit"] / 100
+    split_run_length = graph.run_length * results["% Split"] / 100
+    return graph.run_length - omit_run_length - split_run_length
+
+
 def get_segment_id(filename):
     """
     Gets the segment ID correspionding to the given filename, assuming that