Merge pull request #446 from jamesmartini/develop_regex

Develop to_regex method and associated test

src/cogent3/core/moltype.py

@@ -742,6 +742,15 @@ def to_json(self):
         data = self.to_rich_dict(for_pickle=False)
         return json.dumps(data)
 
+    def to_regex(self, seq):
+        """returns a regex pattern with ambiguities expanded to a character set"""
+        degen_indices = self.get_degenerate_positions(sequence=seq, include_gap=False)
+        seq = list(seq)  # seq can now be modified
+        for index in degen_indices:
+            expanded = self.ambiguities[seq[index]]
+            seq[index] = f"[{''.join(expanded)}]"
+        return "".join(seq)
+
     def gettype(self):
         """Returns type, e.g. 'dna', 'rna', 'protein'. Delete?"""
         return self.label

tests/test_core/test_moltype.py

@@ -257,6 +257,11 @@ def test_init(self):
 class MolTypeTests(TestCase):
     """Tests of the MolType class. Should support same API as old Alphabet."""
 
+    def test_to_regex(self):
+        seq = "ACYGR"
+        regular_expression = DNA.to_regex(seq=seq)
+        self.assertEqual(regular_expression, "AC[CT]G[AG]")
+
     def test_pickling(self):
         pkl = pickle.dumps(DNA)
         got = pickle.loads(pkl)