Fix: Revert broken Copilot suggestions in canonical_transcript()

Two critical bugs were introduced in commit 66607b7:

1. API Misuse: Attempted to pass a list to genome_feature_children(parent: str)
   - Changed: genome_feature_children(parent=transcript_ids, ...)
   - Result: SyntaxError/TypeError in all tests

2. Missing coordinate adjustment: Removed +1 from exon length calculation
   - Changed: (end - start) instead of (end - start + 1)
   - Result: Incorrect transcript length calculations

This fix reverts to the original per-transcript iteration approach while
preserving the critical +1 for 1-based inclusive coordinates.

Test Results:
- All 35 genome_features tests pass (11 canonical + 24 existing)
- All pre-commit checks pass (ruff, black, flake8)
- Zero regressions

Author: kunal-10-cloud

malariagen_data/anoph/genome_features.py

@@ -424,30 +424,21 @@ def canonical_transcript(
 
         # Calculate transcript lengths and find canonical
         debug("Calculating transcript lengths for each transcript")
-        transcript_ids = df_transcripts["ID"].tolist()
-
-        # Get all exon children for all transcripts in a single pass
-        df_children = self.genome_feature_children(parent=transcript_ids, attributes=None)
-        if df_children is None or len(df_children) == 0:
-            raise ValueError(f"Gene '{gene}' has no transcripts with exons")
-
-        # Filter for exons only (important: exclude other feature types)
-        df_exons = df_children[df_children["type"] == "exon"].copy()
-        if len(df_exons) == 0:
-            raise ValueError(f"Gene '{gene}' has no transcripts with exons")
-
-        # Calculate exon lengths and sum per transcript (Parent)
-        df_exons = df_exons.sort_values("start")
-        exon_lengths_series = (df_exons["end"] - df_exons["start"])
-        df_exons = df_exons.assign(_exon_length=exon_lengths_series)
-        exon_length_per_transcript = df_exons.groupby("Parent")["_exon_length"].sum()
-
-        # Build transcript_lengths dict and emit debug for each transcript
         transcript_lengths = {}
-        for transcript_id, exon_length in exon_length_per_transcript.items():
-            transcript_lengths[transcript_id] = exon_length
-            n_exons = (df_exons["Parent"] == transcript_id).sum()
-            debug(f"  {transcript_id}: {n_exons} exons, {exon_length} bp")
+
+        for transcript_id in df_transcripts["ID"]:
+            # Get all exon children (genome_feature_children handles multi-parent exons)
+            df_exons = self.genome_feature_children(
+                parent=transcript_id, attributes=None
+            )
+            # Filter for exons only (important: exclude other feature types)
+            df_exons = df_exons[df_exons["type"] == "exon"].sort_values("start")
+
+            if len(df_exons) > 0:
+                # Calculate total transcribed length (1-based inclusive coordinates)
+                exon_lengths = (df_exons["end"] - df_exons["start"] + 1).sum()
+                transcript_lengths[transcript_id] = exon_lengths
+                debug(f"  {transcript_id}: {len(df_exons)} exons, {exon_lengths} bp")
         if not transcript_lengths:
             raise ValueError(f"Gene '{gene}' has no transcripts with exons")
 

tests/anoph/test_genome_features.py

@@ -338,6 +338,8 @@ def test_canonical_transcript_single_transcript_gene(ag3_sim_api):
 
     if not found_single_transcript_gene:
         pytest.skip("No gene with exactly one transcript available in fixture")
+
+
 def test_canonical_transcript_calculation_correctness(ag3_sim_api):
     """Test that the returned transcript actually has the highest exon length."""
     genes = ag3_sim_api.genome_features().query(
@@ -359,7 +361,7 @@ def test_canonical_transcript_calculation_correctness(ag3_sim_api):
     for transcript_id in all_transcripts["ID"]:
         exons = ag3_sim_api.genome_feature_children(parent=transcript_id)
         exons = exons[exons["type"] == "exon"]
-        length = (exons["end"] - exons["start"]).sum()
+        length = (exons["end"] - exons["start"] + 1).sum()
         if length > max_length:
             max_length = length
             max_transcript = transcript_id
@@ -370,7 +372,7 @@ def test_canonical_transcript_calculation_correctness(ag3_sim_api):
     # Verify canonical has the correct length
     canonical_exons = ag3_sim_api.genome_feature_children(parent=canonical)
     canonical_exons = canonical_exons[canonical_exons["type"] == "exon"]
-    canonical_length = (canonical_exons["end"] - canonical_exons["start"]).sum()
+    canonical_length = (canonical_exons["end"] - canonical_exons["start"] + 1).sum()
     assert canonical_length == max_length