GFF3 simplifier refactoring and tests

src/python/ensembl/io/genomio/gff3/exceptions.py

@@ -16,8 +16,22 @@
 
 __all__ = [
     "GFFParserError",
+    "IgnoredFeatureError",
+    "UnsupportedFeatureError",
 ]
 
 
 class GFFParserError(Exception):
     """Error when parsing a GFF3 file."""
+
+    def __init__(self, message):
+        super().__init__(message)
+        self.message = message
+
+
+class IgnoredFeatureError(GFFParserError):
+    """GFF3 feature can be ignored."""
+
+
+class UnsupportedFeatureError(GFFParserError):
+    """GFF3 feature is not supported."""

src/python/ensembl/io/genomio/gff3/extract_annotation.py

@@ -298,13 +298,10 @@ def store_gene(self, gene: SeqFeature) -> None:
         """Record the functional_annotations of a gene and its children features."""
         self.add_feature(gene, "gene")
 
-        cds_found = False
         for transcript in gene.sub_features:
             self.add_feature(transcript, "transcript", gene.id)
             for feat in transcript.sub_features:
-                if feat.type != "CDS":
-                    continue
-                # Store CDS functional annotation only once
-                if not cds_found:
-                    cds_found = True
+                if feat.type == "CDS":
                     self.add_feature(feat, "translation", transcript.id)
+                    # Store CDS functional annotation only once
+                    break

src/python/ensembl/io/genomio/gff3/restructure.py

@@ -22,6 +22,7 @@
     "move_only_exons_to_new_mrna",
     "move_cds_to_existing_mrna",
     "remove_extra_exons",
+    "remove_cds_from_pseudogene",
 ]
 
 from collections import Counter
@@ -264,3 +265,28 @@ def remove_extra_exons(gene: SeqFeature) -> None:
             gene.sub_features += others
         else:
             raise GFFParserError(f"Can't remove extra exons for {gene.id}, not all start with 'id-'")
+
+
+def remove_cds_from_pseudogene(gene: SeqFeature) -> None:
+    """Removes the CDSs from a pseudogene.
+
+    This assumes the CDSs are sub features of the transcript or the gene.
+
+    """
+    if gene.type != "pseudogene":
+        return
+
+    gene_subfeats = []
+    for transcript in gene.sub_features:
+        if transcript.type == "CDS":
+            logging.debug(f"Remove pseudo CDS {transcript.id}")
+        else:
+            new_subfeats = []
+            for feat in transcript.sub_features:
+                if feat.type == "CDS":
+                    logging.debug(f"Remove pseudo CDS {feat.id}")
+                else:
+                    new_subfeats.append(feat)
+            transcript.sub_features = new_subfeats
+            gene_subfeats.append(transcript)
+    gene.sub_features = gene_subfeats