Merge pull request #9 from VariantEffect/targetseq-fix

fix amino acid target seq validation

mavehgvs/variant.py

@@ -2,12 +2,17 @@
 import itertools
 from typing import Optional, Union, List, Tuple, Mapping, Any, Sequence, Dict, Generator
 
+from fqfa.constants import AA_CODES
+
 from mavehgvs.position import VariantPosition
 from mavehgvs.patterns.combined import any_variant
 from mavehgvs.exceptions import MaveHgvsParseError
 
 __all__ = ["Variant"]
 
+AA_3_TO_1 = {value: key for key, value in AA_CODES.items()}
+"""Dict[str, str]: for converting three-letter amino acid codes to single-letter codes.
+"""
 
 class Variant:
     fullmatch = re.compile(any_variant, flags=re.ASCII).fullmatch
@@ -178,7 +183,11 @@ def sort_key(x):
         if targetseq is not None:
             for vtype, pos, seq in self.variant_tuples():
                 if vtype == "sub":
-                    self._target_validate_substitution(pos, seq[0], targetseq)
+                    if self._prefix == "p":
+                        ref = AA_3_TO_1[seq[0]]
+                    else:
+                        ref = seq[0]
+                    self._target_validate_substitution(pos, ref, targetseq)
                 elif vtype in ("ins", "del", "dup", "delins"):
                     self._target_validate_indel(pos, targetseq)
 

setup.py

@@ -11,7 +11,7 @@
 
 setuptools.setup(
     name="mavehgvs",
-    version="0.2.0",
+    version="0.2.1",
     author="Daniel Esposito and Alan F Rubin",
     author_email="alan.rubin@wehi.edu.au",
     description=(

tests/test_variant.py

@@ -497,6 +497,30 @@ def test_invalid_dna_delins(self):
                 with self.assertRaises(MaveHgvsParseError):
                     Variant(s, targetseq=target)
 
+    def test_matching_protein_substitution(self):
+        variant_tuples = [
+            ("RCQY", "p.Arg1Ala"),
+            ("RCQY", "p.Gln3Trp"),
+            ("RCQY", "p.[Arg1Ala;Gln3Trp]"),
+        ]
+
+        for target, s in variant_tuples:
+            with self.subTest(target=target, s=s):
+                v = Variant(s, targetseq=target)
+                self.assertEqual(s, str(v))
+
+    def test_nonmatching_protein_substitution(self):
+        variant_tuples = [
+            ("RCQY", "p.Cys1Ala"),
+            ("RCQY", "p.Ala3Trp"),
+            ("RCQY", "p.[Arg1Ala;Cys3Trp]"),
+        ]
+
+        for target, s in variant_tuples:
+            with self.subTest(target=target, s=s):
+                with self.assertRaises(MaveHgvsParseError):
+                    Variant(s, targetseq=target)
+
 
 class TestMiscMethods(unittest.TestCase):
     def test_is_multi_variant(self):