WrongLetterError

Raise if sequence and alphabet letters do not match
Edinburgh-Genome-Foundry · Oct 22, 2020 · a003439 · a003439
1 parent 0031b49
commit a003439
Show file tree

Hide file tree

Showing 2 changed files with 16 additions and 0 deletions.
diff --git a/polymera/polymera.py b/polymera/polymera.py
@@ -1,6 +1,10 @@
 import math
 
 
+class WrongLetterError(ValueError):
+    pass
+
+
 class Polymer:
     """Class for representing a sequence with its alphabet.
 
@@ -15,6 +19,13 @@ class Polymer:
     """
 
     def __init__(self, sequence, alphabet):
+        letters_only = set(sequence.to_string()) - set(sequence.separators.values())
+        wrong_letters = letters_only - alphabet.letters
+        if wrong_letters != set():
+            raise WrongLetterError(
+                "These letters are not in the alphabet: %s" % wrong_letters
+            )
+
         self.sequence = sequence
         self.alphabet = alphabet
 

diff --git a/tests/test_polymera.py b/tests/test_polymera.py
@@ -21,6 +21,11 @@ def test_alphabet():
 
 
 def test_polymer():
+    sequence = polymera.Sequence()
+    sequence.add_sequence_from_string("ATGCX")
+    with pytest.raises(ValueError):
+        polymera.Polymer(sequence, alphabet=polymera.bio.DNAAlphabet)
+
     sequence = polymera.Sequence()
     sequence.add_sequence_from_string("ATGAA,ATGCC|TATATTAGAAAAAA")
     sequence.add_sequence_from_string("ATGAA,ATGCC")