Fixed doc string for sage to internal.

spectrum_fundamentals/mod_string.py

@@ -3,40 +3,50 @@
 from itertools import repeat
 from typing import Dict, List, Optional, Tuple
 
-from .constants import MAXQUANT_VAR_MODS, MOD_MASSES, MOD_NAMES, SPECTRONAUT_MODS , MOD_MASSES_SAGE
+from .constants import MAXQUANT_VAR_MODS, MOD_MASSES, MOD_MASSES_SAGE, MOD_NAMES, SPECTRONAUT_MODS
 
-def sage_to_internal(strings: List[str]) -> List[str]:
+
+def sage_to_internal(sequences: List[str]) -> List[str]:
+    """
+    Convert mod string from sage to the internal format.
+
+    This function converts sequences using the mass change of a modification in
+    square brackets as done by Sage to the internal format by replacing the mass
+    shift with the corresponding UNIMOD identifier of known and supported
+    modifications defined in the constants.
+
+    :param sequences: A list of sequences with values inside square brackets.
+    :return: A list of modified sequences with values converted to internal format.
+    """
+    # Define a regular expression pattern to match values within square brackets, like [+1.0] or [-2.0].
+    pattern = r"\[([\+\-]\d+\.\d+)\]"
+
+    # Define a function 'replace' that takes a regex match object.
+    def replace(match):
+        # Extract the value inside the square brackets as a float.
+        value = float(match.group(1))
+
+        # Check if the 'MOD_MASSES_SAGE' dictionary has a replacement value for the extracted value.
+        # If it does, use the replacement value; otherwise, use the original value from the match.
+        unimod_expression = MOD_MASSES_SAGE.get(value, match.group(0))
+
+        return unimod_expression
+
+    # Create an empty list 'modified_strings' to store the modified sequences.
     modified_strings = []
 
-    for string in strings:
-        # Find the number within square brackets (as a float)
-        start_idx = string.find('[') + 1
-        end_idx = string.find(']')
-
-        if start_idx > 0 and end_idx > start_idx:
-            # Extract the number string
-            number_str = string[start_idx:end_idx]
-
-            try:
-                # Attempt to convert the number to a float
-                number = float(number_str)
-            except ValueError:
-                # If conversion fails, keep the original text
-                modified_strings.append(string)
-                continue
-
-            # Replace with the corresponding value from the dictionary
-            if number in MOD_MASSES_SAGE:
-                modified_value = string.replace(f'[{number_str}]', MOD_MASSES_SAGE[number])
-                modified_strings.append(modified_value)
-            else:
-                modified_strings.append(string)
-        else:
-            modified_strings.append(string)
+    # Iterate through the input 'sequences'.
+    for string in sequences:
+        # Use 're.sub' to search and replace values within square brackets in the 'string' using the 'replace' function.
+        modified_string = re.sub(pattern, replace, string)
 
+        # Append the modified string to the 'modified_strings' list.
+        modified_strings.append(modified_string)
+
+    # Return the list of modified sequences.
     return modified_strings
 
-    
+
 def internal_to_spectronaut(sequences: List[str]) -> List[str]:
     """
     Function to translate a modstring from the internal format to the spectronaut format.
@@ -270,4 +280,3 @@ def get_mods_list(mods_variable: str, mods_fixed: str):
         return mods_variable.split(";")
     else:
         return mods_variable.split(";") + mods_fixed.split(";")
-