Merge pull request #242 from daichengxin/master

Fixed Unimod match bugs

Author: ypriverol

sdrf_pipelines/__init__.py

@@ -1 +1 @@
-__version__ = "0.0.32"
+__version__ = "0.0.33"

sdrf_pipelines/maxquant/maxquant.py

@@ -700,8 +700,9 @@ def maxquant_ify_mods(self, sdrf_mods, mqconfdir):
             else:
                 warning_message = "modification is not supported in MaxQuant. skip " + m
                 self.warnings[warning_message] = self.warnings.get(warning_message, 0) + 1
+        oms_mods = ",".join(oms_mods)
 
-        return ",".join(oms_mods)
+        return oms_mods
 
     def maxquant_convert(
         self,

sdrf_pipelines/openms/openms.py

@@ -206,21 +206,32 @@ def __init__(self) -> None:
         self.silac3 = {"silac light": 1, "silac medium": 2, "silac heavy": 3}
         self.silac2 = {"silac light": 1, "silac heavy": 2}
 
+    def _extract_modification_name(self, mod_string):
+        name_match = re.search("NT=(.+?)(;|$)", mod_string)
+        if name_match:
+            name = name_match.group(1)
+        else:
+            raise ValueError(f"Invalid modification string format (missing NT=): {mod_string}")
+        accession = re.search("AC=(.+?)(;|$)", mod_string)
+        if accession:
+            ptm = self._unimod_database.get_by_accession(accession.group(1))
+        else:
+            ptm = None
+
+        if ptm is None:
+            ptm = self._unimod_database.get_by_name(name)
+
+        if ptm is None:
+            raise ValueError("only UNIMOD modifications supported. " + mod_string)
+        else:
+            return ptm.get_name()
+
     # convert modifications in sdrf file to OpenMS notation
     def openms_ify_mods(self, sdrf_mods):
         oms_mods = []
 
         for m in sdrf_mods:
-            if "AC=UNIMOD" not in m and "AC=Unimod" not in m:
-                raise Exception("only UNIMOD modifications supported. " + m)
-
-            name = re.search("NT=(.+?)(;|$)", m).group(1)
-            name = name.capitalize()
-
-            accession = re.search("AC=(.+?)(;|$)", m).group(1)
-            ptm = self._unimod_database.get_by_accession(accession)
-            if ptm is not None:
-                name = ptm.get_name()
+            name = self._extract_modification_name(m)
 
             # workaround for missing PP in some sdrf TODO: fix in sdrf spec?
             if re.search("PP=(.+?)(;|$)", m) is None:
@@ -1130,8 +1141,9 @@ def save_search_settings_to_file(self, output_filename, sdrf, f2c):
                     label = "itraq8plex"
                 else:
                     label = "itraq4plex"
+
                 # add default ITRAQ modification when sdrf with label not contains ITRAQ modification
-                if "ITRAQ" not in f2c.file2mods[raw][0] and "ITRAQ" not in f2c.file2mods[raw][1]:
+                if "itraq" not in f2c.file2mods[raw][0].lower() and "itraq" not in f2c.file2mods[raw][1].lower():
                     warning_message = (
                         "The sdrf with ITRAQ label doesn't contain label modification. Adding default "
                         "variable modifications."

sdrf_pipelines/openms/unimod.py

@@ -128,3 +128,9 @@ def get_by_accession(self, accession):
             if mod.get_accession() == accession:
                 return mod
         return None
+
+    def get_by_name(self, name):
+        for mod in self.modifications:
+            if mod.get_name().lower() == name.lower():
+                return mod
+        return None