added a few lines for sage

wilhelm-lab · Oct 27, 2023 · df44fb1 · df44fb1
1 parent c8ce0d1
commit df44fb1
Show file tree

Hide file tree

Showing 2 changed files with 40 additions and 30 deletions.
diff --git a/spectrum_fundamentals/constants.py b/spectrum_fundamentals/constants.py
@@ -153,17 +153,17 @@
     "[UNIMOD:35]": 15.9949146,  # Oxidation
     "[UNIMOD:1]": 42.010565,  # Acetylation
 }
-MOD_MASSES_SAGE = {+229.1629: '[UNIMOD:737]', 
-                       +304.2071: '[UNIMOD:2016]', 
-                       +144.1020: '[UNIMOD:214]', 
-                       +304.2053: '[UNIMOD:730]', 
-                       +8.0141: '[UNIMOD:259]', 
-                       +10.0082: '[UNIMOD:267]', 
-                       +79.9663: '[UNIMOD:21]', 
+MOD_MASSES_SAGE = {229.1629: '[UNIMOD:737]', 
+                       304.2071: '[UNIMOD:2016]', 
+                       144.1020: '[UNIMOD:214]', 
+                       304.2053: '[UNIMOD:730]', 
+                       8.0141: '[UNIMOD:259]', 
+                       10.0082: '[UNIMOD:267]', 
+                       79.9663: '[UNIMOD:21]', 
                        -18.0105: '[UNIMOD:23]', 
-                       +57.0214: '[UNIMOD:4]', 
-                       +15.9949: '[UNIMOD:35]', 
-                       +42.0105: '[UNIMOD:1]'}
+                       57.0214: '[UNIMOD:4]', 
+                       15.9949: '[UNIMOD:35]', 
+                       42.0105: '[UNIMOD:1]'}
 # these are only used for prosit_grpc, oktoberfest uses the masses from MOD_MASSES
 AA_MOD_MASSES = {
     "K[UNIMOD:737]": AA_MASSES["K"] + MOD_MASSES["[UNIMOD:737]"],

diff --git a/spectrum_fundamentals/mod_string.py b/spectrum_fundamentals/mod_string.py
@@ -5,26 +5,36 @@
 
 from .constants import MAXQUANT_VAR_MODS, MOD_MASSES, MOD_NAMES, SPECTRONAUT_MODS , MOD_MASSES_SAGE
 
-def sage_to_internal(sequences: List[str])->List[str]:
-    # Find the number within square brackets (as a float)
-    start_idx = sequences.find('[') + 1
-    end_idx = sequences.find(']')
-
-    # Extract the number string
-    number_str = sequences[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
-        return sequences
-
-    # Replace with the corresponding value from the dictionary
-    if number in MOD_MASSES_SAGE:
-        return sequences.replace(f'[{number_str}]', MOD_MASSES_SAGE[number])
-    else:
-        return sequences
+def sage_to_internal(strings: List[str]) -> List[str]:
+    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)
+
+    return modified_strings
 
 
 def internal_to_spectronaut(sequences: List[str]) -> List[str]: