Merge pull request #54 from Janelia-Trackathon-2023/intertrack_edges

Annotate intertrack edges in the ctc `get_edge_errors` function

src/traccuracy/_tracking_graph.py

@@ -385,6 +385,14 @@ def get_divisions(self):
         """
         return [node for node, degree in self.graph.out_degree() if degree >= 2]
 
+    def get_merges(self):
+        """Get all nodes that have at least two incoming edges from the previous time frame
+
+        Returns:
+            list of hashable: a list of node ids for nodes that have more than one parent
+        """
+        return [node for node, degree in self.graph.in_degree() if degree >= 2]
+
     def get_preds(self, node):
         """Get all predecessors of the given node.
 

src/traccuracy/loaders/_ctc.py

@@ -113,7 +113,6 @@ def ctc_to_graph(df, detections):
                 {
                     "source": cellids[0:-1],
                     "target": cellids[1:],
-                    "is_intertrack_edge": [0 for _ in range(len(cellids) - 1)],
                 }
             )
         )
@@ -126,11 +125,7 @@ def ctc_to_graph(df, detections):
 
         target = "{}_{}".format(row["Cell_ID"], row["Start"])
 
-        edges.append(
-            pd.DataFrame(
-                {"source": [source], "target": [target], "is_intertrack_edge": [1]}
-            )
-        )
+        edges.append(pd.DataFrame({"source": [source], "target": [target]}))
 
     # Store position attributes on nodes
     detections["node_id"] = (
@@ -149,9 +144,8 @@ def ctc_to_graph(df, detections):
 
     # Create graph
     edges = pd.concat(edges)
-    edges["is_intertrack_edge"] = edges["is_intertrack_edge"].astype(bool)
     G = nx.from_pandas_edgelist(
-        edges, source="source", target="target", create_using=nx.DiGraph, edge_attr=True
+        edges, source="source", target="target", create_using=nx.DiGraph
     )
 
     # Add all isolates to graph

src/traccuracy/track_errors/_ctc.py

@@ -80,12 +80,16 @@ def get_edge_errors(matched_data: "Matched"):
         logger.info("Edge errors already calculated. Skipping graph annotation")
         return
 
+    # Node errors must already be annotated
+    if not comp_graph.node_errors and not gt_graph.node_errors:
+        logger.warning("Node errors have not been annotated. Running node annotation.")
+        get_vertex_errors(matched_data)
+
     induced_graph = comp_graph.get_subgraph(
         comp_graph.get_nodes_with_flag(NodeAttr.TRUE_POS)
     ).graph
 
     comp_graph.set_edge_attribute(list(comp_graph.edges()), EdgeAttr.FALSE_POS, False)
-    comp_graph.set_edge_attribute(list(comp_graph.edges()), EdgeAttr.TRUE_POS, False)
     comp_graph.set_edge_attribute(
         list(comp_graph.edges()), EdgeAttr.WRONG_SEMANTIC, False
     )
@@ -94,6 +98,23 @@ def get_edge_errors(matched_data: "Matched"):
     node_mapping_first = np.array([mp[0] for mp in node_mapping])
     node_mapping_second = np.array([mp[1] for mp in node_mapping])
 
+    # intertrack edges = connection between parent and daughter
+    for graph in [comp_graph, gt_graph]:
+        # Set to False by default
+        graph.set_edge_attribute(list(graph.edges()), EdgeAttr.INTERTRACK_EDGE, False)
+
+        for parent in graph.get_divisions():
+            for daughter in graph.get_succs(parent):
+                graph.set_edge_attribute(
+                    (parent, daughter), EdgeAttr.INTERTRACK_EDGE, True
+                )
+
+        for merge in graph.get_merges():
+            for parent in graph.get_preds(merge):
+                graph.set_edge_attribute(
+                    (parent, merge), EdgeAttr.INTERTRACK_EDGE, True
+                )
+
     # fp edges - edges in induced_graph that aren't in gt_graph
     for edge in tqdm(induced_graph.edges, "Evaluating FP edges"):
         source, target = edge[0], edge[1]
@@ -110,8 +131,6 @@ def get_edge_errors(matched_data: "Matched"):
             is_parent_comp = comp_graph.edges()[edge][EdgeAttr.INTERTRACK_EDGE]
             if is_parent_gt != is_parent_comp:
                 comp_graph.set_edge_attribute(edge, EdgeAttr.WRONG_SEMANTIC, True)
-            else:
-                comp_graph.set_edge_attribute(edge, EdgeAttr.TRUE_POS, True)
 
     # fn edges - edges in gt_graph that aren't in induced graph
     for edge in tqdm(gt_graph.edges(), "Evaluating FN edges"):

tests/metrics/test_ctc_metrics.py

@@ -1,5 +1,3 @@
-import networkx as nx
-from traccuracy._tracking_graph import EdgeAttr
 from traccuracy.matchers._ctc import CTCMatched
 from traccuracy.metrics._ctc import CTCMetrics
 
@@ -11,7 +9,6 @@ def test_compute_mapping():
     n_frames = 3
     n_labels = 3
     track_graph = get_movie_with_graph(ndims=3, n_frames=n_frames, n_labels=n_labels)
-    nx.set_edge_attributes(track_graph.graph, 0, EdgeAttr.INTERTRACK_EDGE)
 
     matched = CTCMatched(gt_graph=track_graph, pred_graph=track_graph)
     metric = CTCMetrics(matched)

tests/test_tracking_graph.py

@@ -65,6 +65,40 @@ def nx_comp2():
     return graph
 
 
+@pytest.fixture
+def nx_merge():
+    """
+    3_0--3_1--\\
+              3_2--3_3
+    3_4--3_5--/
+    """
+    cells = [
+        {"id": "3_0", "t": 0, "x": 0, "y": 0},
+        {"id": "3_1", "t": 1, "x": 0, "y": 0},
+        {"id": "3_2", "t": 2, "x": 0, "y": 0},
+        {"id": "3_3", "t": 3, "x": 0, "y": 0},
+        {"id": "3_4", "t": 0, "x": 0, "y": 0},
+        {"id": "3_5", "t": 1, "x": 0, "y": 0},
+    ]
+
+    edges = [
+        {"source": "3_0", "target": "3_1"},
+        {"source": "3_1", "target": "3_2"},
+        {"source": "3_2", "target": "3_3"},
+        {"source": "3_4", "target": "3_5"},
+        {"source": "3_5", "target": "3_2"},
+    ]
+    graph = nx.DiGraph()
+    graph.add_nodes_from([(cell["id"], cell) for cell in cells])
+    graph.add_edges_from([(edge["source"], edge["target"]) for edge in edges])
+    return graph
+
+
+@pytest.fixture
+def merge_graph(nx_merge):
+    return TrackingGraph(nx_merge)
+
+
 @pytest.fixture
 def simple_graph(nx_comp1):
     return TrackingGraph(nx_comp1)
@@ -168,11 +202,20 @@ def test_get_divisions(complex_graph):
     assert complex_graph.get_divisions() == ["1_1", "2_2"]
 
 
-def test_get_preds(simple_graph):
+def test_get_merges(merge_graph):
+    assert merge_graph.get_merges() == ["3_2"]
+
+
+def test_get_preds(simple_graph, merge_graph):
+    # Division graph
     assert simple_graph.get_preds("1_0") == []
     assert simple_graph.get_preds("1_1") == ["1_0"]
     assert simple_graph.get_preds("1_2") == ["1_1"]
 
+    # Merge graph
+    assert merge_graph.get_preds("3_3") == ["3_2"]
+    assert merge_graph.get_preds("3_2") == ["3_1", "3_5"]
+
 
 def test_get_succs(simple_graph):
     assert simple_graph.get_succs("1_0") == ["1_1"]

tests/track_errors/test_ctc_errors.py

@@ -79,7 +79,6 @@ def test_assign_edge_errors():
     comp_g.add_nodes_from(comp_ids)
     comp_g.add_edges_from(comp_edges)
     nx.set_node_attributes(comp_g, True, NodeAttr.TRUE_POS)
-    nx.set_edge_attributes(comp_g, 0, EdgeAttr.INTERTRACK_EDGE)
     nx.set_node_attributes(
         comp_g,
         {idx: {"t": 0, "segmentation_id": 1, "y": 0, "x": 0} for idx in comp_ids},
@@ -90,7 +89,6 @@ def test_assign_edge_errors():
     gt_g = nx.DiGraph()
     gt_g.add_nodes_from(gt_ids)
     gt_g.add_edges_from(gt_edges)
-    nx.set_edge_attributes(gt_g, 0, EdgeAttr.INTERTRACK_EDGE)
     nx.set_node_attributes(gt_g, False, NodeAttr.FALSE_NEG)
     nx.set_node_attributes(
         gt_g, {idx: {"t": 0, "segmentation_id": 1, "y": 0, "x": 0} for idx in gt_ids}
@@ -102,51 +100,45 @@ def test_assign_edge_errors():
 
     get_edge_errors(matched_data)
 
-    assert comp_g.edges[(3, 4)][EdgeAttr.TRUE_POS]
     assert comp_g.edges[(7, 8)][EdgeAttr.FALSE_POS]
     assert gt_g.edges[(17, 18)][EdgeAttr.FALSE_NEG]
 
 
 def test_assign_edge_errors_semantics():
-    comp_ids = [3, 7, 10]
-    comp_ids_2 = list(np.asarray(comp_ids) + 1)
-    comp_ids += comp_ids_2
-
-    gt_ids = [4, 12, 17]
-    gt_ids_2 = list(np.asarray(gt_ids) + 1)
-    gt_ids += gt_ids_2
-
-    mapping = [(4, 3), (12, 7), (17, 10), (5, 4), (18, 11), (13, 8)]
-
-    # need a tp, fp, fn
-    comp_edges = [(3, 4)]
-    comp_g = nx.DiGraph()
-    comp_g.add_nodes_from(comp_ids)
-    comp_g.add_edges_from(comp_edges)
-    nx.set_node_attributes(comp_g, True, NodeAttr.TRUE_POS)
-    nx.set_edge_attributes(comp_g, 0, EdgeAttr.INTERTRACK_EDGE)
-    nx.set_node_attributes(
-        comp_g,
-        {idx: {"t": 0, "segmentation_id": 1, "y": 0, "x": 0} for idx in comp_ids},
-    )
-    G_comp = TrackingGraph(comp_g)
-
-    gt_edges = [(4, 5), (17, 18)]
-    gt_g = nx.DiGraph()
-    gt_g.add_nodes_from(gt_ids)
-    gt_g.add_edges_from(gt_edges)
-    nx.set_edge_attributes(gt_g, 0, EdgeAttr.INTERTRACK_EDGE)
-    nx.set_node_attributes(gt_g, False, NodeAttr.FALSE_NEG)
-    gt_g.edges[(4, 5)][EdgeAttr.INTERTRACK_EDGE] = 1
-    nx.set_node_attributes(
-        gt_g, {idx: {"t": 0, "segmentation_id": 1, "y": 0, "x": 0} for idx in gt_ids}
-    )
-    G_gt = TrackingGraph(gt_g)
-
-    matched_data = DummyMatched(G_gt, G_comp)
+    """
+    gt:
+    1_0 -- 1_1 -- 1_2 -- 1_3
+
+    comp:
+                         1_3
+    1_0 -- 1_1 -- 1_2 -<
+                         2_3
+    """
+
+    gt = nx.DiGraph()
+    gt.add_edge("1_0", "1_1")
+    gt.add_edge("1_1", "1_2")
+    gt.add_edge("1_2", "1_3")
+    # Set node attrs
+    attrs = {}
+    for node in gt.nodes:
+        attrs[node] = {"t": int(node[-1:]), "x": 0, "y": 0}
+    nx.set_node_attributes(gt, attrs)
+
+    comp = gt.copy()
+    comp.add_edge("1_2", "2_3")
+    # Set node attrs
+    attrs = {}
+    for node in comp.nodes:
+        attrs[node] = {"t": int(node[-1:]), "x": 0, "y": 0}
+    nx.set_node_attributes(comp, attrs)
+
+    # Define mapping with all nodes matching except for 2_3 in comp
+    mapping = [(n, n) for n in gt.nodes]
+
+    matched_data = DummyMatched(TrackingGraph(gt), TrackingGraph(comp))
     matched_data.mapping = mapping
 
     get_edge_errors(matched_data)
 
-    assert comp_g.edges[(3, 4)][EdgeAttr.WRONG_SEMANTIC]
-    assert not comp_g.edges[(3, 4)][EdgeAttr.TRUE_POS]
+    assert comp.edges[("1_2", "1_3")][EdgeAttr.WRONG_SEMANTIC]