Merge pull request #76 from wilhelm-lab/fix/tmt_modstring

Fix/tmt modstring
wilhelm-lab · Sep 19, 2023 · cbf31c8 · cbf31c8
2 parents 86695a3 + bcc1355
commit cbf31c8
Show file tree

Hide file tree

Showing 6 changed files with 601 additions and 626 deletions.
diff --git a/poetry.lock b/poetry.lock
diff --git a/pyproject.toml b/pyproject.toml
@@ -20,7 +20,7 @@ classifiers = [
 
 
 [tool.poetry.dependencies]
-python = "^3.8.0"
+python = ">=3.8.0,<3.11.0"
 click = ">=8.0.0"
 rich = ">=10.3.0"
 PyYAML = ">=5.4.1"

diff --git a/spectrum_fundamentals/constants.py b/spectrum_fundamentals/constants.py
@@ -49,7 +49,6 @@
     "C[UNIMOD:4]": 2,
     "K[UNIMOD:737]": 22,
     "K[UNIMOD:2016]": 22,
-    "K[UNIMOD:2016]": 22,
     "K[UNIMOD:214]": 22,
     "K[UNIMOD:730]": 22,
     "S[UNIMOD:21]": 25,
@@ -73,16 +72,16 @@
     "(ox)": "[UNIMOD:35]",
     "(Oxidation (M))": "[UNIMOD:35]",
     "(tm)": "[UNIMOD:737]",
-    "_(tm)": "_[UNIMOD:737]",
+    "_(tm)": "_[UNIMOD:737]-",
     "K(tm)": "K[UNIMOD:737]",
     "(i4)": "[UNIMOD:214]",
-    "_(i4)": "_[UNIMOD:214]",
+    "_(i4)": "_[UNIMOD:214]-",
     "K(i4)": "K[UNIMOD:214]",
     "(i8)": "[UNIMOD:730]",
-    "_(i8)": "_[UNIMOD:730]",
+    "_(i8)": "_[UNIMOD:730]-",
     "K(i8)": "K[UNIMOD:730]",
     "(tmp)": "[UNIMOD:2016]",
-    "_(tmp)": "_[UNIMOD:2016]",
+    "_(tmp)": "_[UNIMOD:2016]-",
     "K(tmp)": "K[UNIMOD:2016]",
     "(ph)": "[UNIMOD:21]",
     "(Phospho (STY))": "[UNIMOD:21]",

diff --git a/spectrum_fundamentals/fragments.py b/spectrum_fundamentals/fragments.py
@@ -24,10 +24,11 @@ def _get_modifications(peptide_sequence: str) -> Optional[Tuple[Dict[int, float]
     modification_mass = constants.MOD_MASSES
     # Handle terminal modifications here
     for possible_tmt_mod in constants.TMT_MODS.values():
-        if peptide_sequence.startswith(possible_tmt_mod):  # TMT_6
+        n_term_tmt = possible_tmt_mod + "-"
+        if peptide_sequence.startswith(n_term_tmt):
             tmt_n_term = 2
             modification_deltas.update({0: constants.MOD_MASSES[possible_tmt_mod]})
-            peptide_sequence = peptide_sequence[len(possible_tmt_mod) :]
+            peptide_sequence = peptide_sequence[len(n_term_tmt) :]
             break
 
     if "(" in peptide_sequence:
@@ -183,7 +184,7 @@ def initialize_peaks(
             for ion_type in range(0, number_of_ion_types):  # generate all ion types
                 # Check for neutral loss here
                 mass = (ion_type_masses[ion_type] + charge_delta) / charge
-                min_mass, max_mass = get_min_max_mass(mass_analyzer, mass)
+                min_mass, max_mass = get_min_max_mass(mass_analyzer, mass, mass_tolerance, unit_mass_tolerance)
                 fragments_meta_data.append(
                     {
                         "ion_type": ion_types[ion_type],  # ion type

diff --git a/tests/unit_tests/test_fragments.py b/tests/unit_tests/test_fragments.py
@@ -18,11 +18,11 @@ def test_get_modifications_carbamidomethylation(self):
 
     def test_get_modifications_tmt_tag(self):
         """Test get_modifications."""
-        assert fragments._get_modifications("[UNIMOD:737]ABC[UNIMOD:4]") == ({0: 229.162932, 2: 57.02146}, 2, "ABC")
+        assert fragments._get_modifications("[UNIMOD:737]-ABC[UNIMOD:4]") == ({0: 229.162932, 2: 57.02146}, 2, "ABC")
 
     def test_get_modifications_tmtpro_tag(self):
         """Test get_modifications."""
-        assert fragments._get_modifications("[UNIMOD:2016]ABC[UNIMOD:4]") == ({0: 304.207146, 2: 57.02146}, 2, "ABC")
+        assert fragments._get_modifications("[UNIMOD:2016]-ABC[UNIMOD:4]") == ({0: 304.207146, 2: 57.02146}, 2, "ABC")
 
 
 class TestComputeMasses(unittest.TestCase):
@@ -63,6 +63,11 @@ def test_compute_peptide_masses(self):
         seq = "SEQUENC[UNIMOD:4]E"
         self.assertEqual(fragments.compute_peptide_mass(seq), 1045.2561516699998)
 
+    def test_compute_peptide_masses_tmtpro(self):
+        """Test computation of peptide masses with valid input and tmt tag."""
+        seq = "[UNIMOD:737]-SEQUENC[UNIMOD:4]E"
+        self.assertEqual(fragments.compute_peptide_mass(seq), 1274.41908367)
+
     def test_compute_peptide_masses_with_invalid_syntax(self):
         """Negative testing of comuptation of peptide mass with unsupported syntax of mod string."""
         seq = "SEQUEM(Ox.)CE"

diff --git a/tests/unit_tests/test_mod_string.py b/tests/unit_tests/test_mod_string.py
@@ -39,9 +39,9 @@ def test_maxquant_to_internal_variable_dehydration_long(self):
 
     def test_maxquant_to_internal_tmt(self):
         """Test maxquant_to_internal_tmt."""
-        fixed_mods = {"C": "C[UNIMOD:4]", "^_": "_[UNIMOD:737]", "K": "K[UNIMOD:737]"}
+        fixed_mods = {"C": "C[UNIMOD:4]", "^_": "_[UNIMOD:737]-", "K": "K[UNIMOD:737]"}
         self.assertEqual(
-            mod.maxquant_to_internal(["_ABCDEFGHK_"], fixed_mods), ["[UNIMOD:737]ABC[UNIMOD:4]DEFGHK[UNIMOD:737]"]
+            mod.maxquant_to_internal(["_ABCDEFGHK_"], fixed_mods), ["[UNIMOD:737]-ABC[UNIMOD:4]DEFGHK[UNIMOD:737]"]
         )
 
     def test_maxquant_to_internal_silac(self):