Fix protein extension conversion

Author: mihailefter

mutalyzer_hgvs_parser/ebnf/protein.g

@@ -43,7 +43,7 @@ extension: extension_n | extension_c
 
 extension_n:  "ext" "-" NUMBER
 
-extension_c:  P_SEQUENCE "ext" (P_SEQUENCE | p_point)
+extension_c:  P_SEQUENCE "ext" P_SEQUENCE p_location?
 
 frame_shift: "fs" | AA "fs" ("*" | "Ter") p_location
 

mutalyzer_hgvs_parser/hgvs_parser.py

@@ -540,12 +540,16 @@ def extension_n(self, children):
         return Tree("inserted", [Tree("insert", location)])
 
     def extension_c(self, children):
-        inserted = [Tree("insert", [Token("P_SEQUENCE", children[0]).value])]
-        if isinstance(children[1], Token):
-            inserted.append(Tree("insert", [Token("P_SEQUENCE", children[1].value)]))
+        new_children = []
+        for child in children:
+            if isinstance(child, Token):
+                new_children.append(Tree("insert", [Token("P_SEQUENCE", child.value)]))
+            else:
+                new_children.append(Tree("insert", [child]))
+        if new_children:
+            return Tree("extension", [Tree("inserted", new_children)])
         else:
-            inserted.append(Tree("insert", [Tree("location", [children[1]])]))
-        return Tree("inserted", inserted)
+            return Tree("extension", [])
 
     def frame_shift(self, children):
         new_children = []

tests/test_protein.py

@@ -831,10 +831,10 @@
                 "source": "reference",
                 "inserted": [
                     {"sequence": "Gln", "source": "description"},
+                    {"sequence": "Ter", "source": "description"},
                     {
                         "location": {
                             "type": "point",
-                            "amino_acid": "Ter",
                             "position": 17,
                         },
                         "source": "reference",
@@ -854,10 +854,10 @@
                 "source": "reference",
                 "inserted": [
                     {"sequence": "Gln", "source": "description"},
+                    {"sequence": "*", "source": "description"},
                     {
                         "location": {
                             "type": "point",
-                            "amino_acid": "*",
                             "position": 17,
                         },
                         "source": "reference",
@@ -901,8 +901,46 @@
         ],
     },
     # -
-    "PREF:p.Ter327Argext*?": {},
-    "PREF:p.*327Argext*?": {},
+    "PREF:p.Ter327Argext*?": {
+        "type": "description_protein",
+        "reference": {"id": "PREF"},
+        "coordinate_system": "p",
+        "variants": [
+            {
+                "location": {"type": "point", "amino_acid": "Ter", "position": 327},
+                "type": "extension",
+                "source": "reference",
+                "inserted": [
+                    {"sequence": "Arg", "source": "description"},
+                    {"sequence": "*", "source": "description"},
+                    {
+                        "location": {"type": "point", "uncertain": True},
+                        "source": "reference",
+                    },
+                ],
+            }
+        ],
+    },
+    "PREF:p.*327Argext*?": {
+        "type": "description_protein",
+        "reference": {"id": "PREF"},
+        "coordinate_system": "p",
+        "variants": [
+            {
+                "location": {"type": "point", "amino_acid": "*", "position": 327},
+                "type": "extension",
+                "source": "reference",
+                "inserted": [
+                    {"sequence": "Arg", "source": "description"},
+                    {"sequence": "*", "source": "description"},
+                    {
+                        "location": {"type": "point", "uncertain": True},
+                        "source": "reference",
+                    },
+                ],
+            }
+        ],
+    },
 }
 
 OTHER = {