Merge pull request #78 from wilhelm-lab/feature/sage_dict_parse

added sage reversed unimod dictionary

spectrum_fundamentals/constants.py

@@ -168,7 +168,19 @@
     "[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]",
+    -18.0105: "[UNIMOD:23]",
+    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]"],

spectrum_fundamentals/mod_string.py

@@ -3,7 +3,55 @@
 from itertools import repeat
 from typing import Dict, List, Optional, Tuple
 
-from .constants import MAXQUANT_VAR_MODS, MOD_MASSES, MOD_NAMES, MSFRAGGER_VAR_MODS, SPECTRONAUT_MODS
+from .constants import MAXQUANT_VAR_MODS, MOD_MASSES, MOD_MASSES_SAGE, MOD_NAMES, MSFRAGGER_VAR_MODS, SPECTRONAUT_MODS
+
+
+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"[A-Z]?\[([\+\-]\d+\.\d+)\]-?"
+
+    # 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))
+        key = match.string[match.start() : match.end()]
+        if key.endswith("-"):
+            unimod_expression = f"{MOD_MASSES_SAGE.get(value, match.group(0))}-"
+        elif key.startswith("C"):
+            unimod_expression = f"C{MOD_MASSES_SAGE.get(value, match.group(0))}"
+        elif key.startswith("K"):
+            unimod_expression = f"K{MOD_MASSES_SAGE.get(value, match.group(0))}"
+        elif key.startswith("M"):
+            unimod_expression = f"M{MOD_MASSES_SAGE.get(value, match.group(0))}"
+
+        # 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.
+        return unimod_expression
+
+    # Create an empty list 'modified_strings' to store the modified sequences.
+    modified_strings = []
+
+    # 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]:

tests/unit_tests/test_mod_string.py

@@ -18,6 +18,25 @@ def test_internal_to_mod_names(self):
         ]
 
 
+class TestSageToInternal(unittest.TestCase):
+    """Class to test MaxQuant to internal."""
+
+    def test_sage_to_internal_carbamidomethylation(self):
+        """Test maxquant_to_internal_carbamidomethylation."""
+        self.assertEqual(mod.sage_to_internal(["ABC[+57.0214]DEFGH"]), ["ABC[UNIMOD:4]DEFGH"])
+
+    def test_sage_to_internal_variable_oxidation(self):
+        """Test maxquant_to_internal_variable_oxidation."""
+        self.assertEqual(mod.sage_to_internal(["ABC[+57.0214]DM[+15.9949]EFGH"]), ["ABC[UNIMOD:4]DM[UNIMOD:35]EFGH"])
+
+    def test_sage_to_internal_tmt(self):
+        """Test maxquant_to_internal_tmt."""
+        self.assertEqual(
+            mod.sage_to_internal(["[+229.1629]-ABC[+57.0214]DEFGHK[+229.1629]"]),
+            ["[UNIMOD:737]-ABC[UNIMOD:4]DEFGHK[UNIMOD:737]"],
+        )
+
+
 class TestMaxQuantToInternal(unittest.TestCase):
     """Class to test MaxQuant to internal."""