Merge pull request #90 from yutanagano/improve-junction-standardisation

Improve junction standardisation
yutanagano · Jan 12, 2025 · 6e13561 · 6e13561
2 parents 4d21d09 + f91170a
commit 6e13561
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 13 deletions.
diff --git a/src/tidytcells/junction/_standardize.py b/src/tidytcells/junction/_standardize.py
@@ -59,25 +59,25 @@ def standardize(
 
     :return:
         If possible, a standardized version of the input string is returned.
-        If the input string cannot be standardized, the function follows the behaviour as set by ``on_fail``.
+        If the input string cannot be standardized, the function follows the behaviour as set by `on_fail`.
     :rtype:
         Union[str, None]
 
     .. topic:: Example usage
 
         Strings that look like junction sequences will be accepted, and returned in capitalised form.
 
-        >>> tt.junction.standardize("csadaff")
-        'CSADAFF'
+        >>> tt.junction.standardize("csadaf")
+        'CSADAF'
 
-        Strings that are valid amino acid sequences but do not stard and end with the appropriate residues will have a C and an F appended to its beginning and end respectively.
+        Strings that are valid amino acid sequences but do not stard and end with the appropriate residues will have a C and an F appended to its beginning and end as required.
 
-        >>> tt.junction.standardize("sadaf")
-        'CSADAFF'
+        >>> tt.junction.standardize("sada")
+        'CSADAF'
 
-        However, setting ``strict`` to ``True`` will cause these cases to be rejected.
+        However, setting `strict` to ``True`` will cause these cases to be rejected.
 
-        >>> result = tt.junction.standardize("sadaf", strict=True)
+        >>> result = tt.junction.standardize("sada", strict=True)
         Input sadaf was rejected as it is not a valid junction sequence.
         >>> print(result)
         None
@@ -92,11 +92,11 @@ def standardize(
             IF input sequence contains non-amino acid symbols:
                 set standardization status to failed
 
-            IF input sequence does not start with C and end with F:
+            IF input sequence does not start with C and end with W / F:
                 IF strict is set to True:
                     set standardization status to failed
                 ELSE:
-                    add C to the beginning and F to the end of the input sequence
+                    add C to the beginning and F to the end of the input sequence as required
                     set standardization status to successful
             ELSE:
                 set standardization status to successful
@@ -150,10 +150,17 @@ def standardize(
                 logger.warning(
                     f"Failed to standardize {original_input}: not a valid junction sequence."
                 )
+
             if on_fail == "reject":
                 return None
+
             return original_input
-        seq = "C" + seq + "F"
+
+        if not seq.startswith("C"):
+            seq = "C" + seq
+
+        if not JUNCTION_MATCHING_REGEX.match(seq):
+            seq = seq + "F"
 
     return seq
 

diff --git a/tests/test_junction.py b/tests/test_junction.py
@@ -30,8 +30,8 @@ def test_various_rejections(self, seq, caplog):
         (
             ("casqyf", "CASQYF"),
             ("ASQY", "CASQYF"),
-            ("CASQY", "CCASQYF"),
-            ("ASQYF", "CASQYFF"),
+            ("CASQY", "CASQYF"),
+            ("ASQYF", "CASQYF"),
         ),
     )
     def test_various_corrections(self, seq, expected):