incorporate reference molecules

implementations/python/mzpaf/annotation.py

@@ -7,12 +7,15 @@
 import re
 from sys import intern
 from typing import Any, List, Optional, Pattern, Dict, Tuple, Union
+import warnings
 
 try:
     from pyteomics.proforma import ProForma
 except ImportError:
     ProForma = None
 
+from .reference import ReferenceMolecule
+
 JSONDict = Dict[str, Union[List, Dict, int, float, str, bool, None]]
 
 annotation_pattern = re.compile(r"""
@@ -602,22 +605,41 @@ class ReferenceIonAnnotation(IonAnnotationBase):
         The reference identifier.
     """
 
-    __slots__ = ("reference", )
+    __slots__ = ("_reference", "reference_molecule")
 
     series_label = "reference"
     _molecule_description_fields = {
         "reference": "The molecule refernce identifier"
     }
 
+    _reference: str
     reference: str
+    reference_molecule: Optional[ReferenceMolecule]
 
     def __init__(self, series, reference, neutral_losses=None, isotope=None, adducts=None, charge=None,
                  analyte_reference=None, mass_error=None, confidence=None, rest=None, is_auxiliary=None):
         super(ReferenceIonAnnotation, self).__init__(
             series, neutral_losses, isotope, adducts, charge, analyte_reference, mass_error, confidence,
             rest, is_auxiliary)
+        self._reference = None
         self.reference = reference
 
+    @property
+    def reference(self) -> str:
+        return self._reference
+
+    @reference.setter
+    def reference(self, value):
+        self._reference = value
+        if value is not None:
+            try:
+                self.reference_molecule = ReferenceMolecule.get(value)
+            except KeyError:
+                warnings.warn(f"Could not find a reference entry for {value}")
+                self.reference_molecule = None
+        else:
+            self.reference_molecule = None
+
     def _format_ion(self):
         return f"r[{self.reference}]"
 

implementations/python/mzpaf/data/reference_molecules.json

@@ -0,0 +1,254 @@
+{
+  "Hex": {
+    "molecule_type": "monosaccharide",
+    "neutral_mass": 162.0528234185,
+    "chemical_formula": "C6H10O5"
+  },
+  "HexNAc": {
+    "molecule_type": "monosaccharide",
+    "neutral_mass": 203.07937251951,
+    "chemical_formula": "C8H13N1O5"
+  },
+  "dHex": {
+    "molecule_type": "monosaccharide",
+    "neutral_mass": 146.05790879894,
+    "chemical_formula": "C6H10O4"
+  },
+  "NeuAc": {
+    "molecule_type": "monosaccharide",
+    "neutral_mass": 291.09541650647,
+    "chemical_formula": "C11H17N1O8"
+  },
+  "NeuGc": {
+    "molecule_type": "monosaccharide",
+    "neutral_mass": 307.09033112603,
+    "chemical_formula": "C11H17N1O9"
+  },
+  "TMT126": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 126.127726
+  },
+  "TMT127N": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 127.124761
+  },
+  "TMT127C": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 127.131081
+  },
+  "TMT128N": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 128.128116
+  },
+  "TMT128C": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 128.134436
+  },
+  "TMT129N": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 129.131471
+  },
+  "TMT129C": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 129.13779
+  },
+  "TMT130N": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 130.134825
+  },
+  "TMT130C": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 130.141145
+  },
+  "TMT131N": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 131.13818
+  },
+  "TMT131C": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 131.1445
+  },
+  "TMT132N": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 132.141535
+  },
+  "TMT132C": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 122.147855
+  },
+  "TMT133N": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 133.14489
+  },
+  "TMT133C": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 133.15121
+  },
+  "TMT134N": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 134.148245
+  },
+  "TMT134C": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 134.154565
+  },
+  "TMT135N": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 135.1516
+  },
+  "TMTzero": {
+    "label_type": "TMTzero",
+    "molecule_type": "reporter+balance",
+    "neutral_mass": 224.152478,
+    "ion_mz": 225.15975447
+  },
+  "TMTpro_zero": {
+    "label_type": "TMTpro_zero",
+    "molecule_type": "reporter+balance",
+    "neutral_mass": 295.189592,
+    "ion_mz": 296.1968685
+  },
+  "TMT2plex": {
+    "label_type": "TMT2plex",
+    "molecule_type": "reporter+balance",
+    "neutral_mass": 225.155833,
+    "ion_mz": 226.16310947
+  },
+  "TMT6plex": {
+    "label_type": "TMT6plex",
+    "molecule_type": "reporter+balance",
+    "neutral_mass": 229.162932,
+    "ion_mz": 230.17020847
+  },
+  "TMTpro": {
+    "label_type": "TMTpro",
+    "molecule_type": "reporter+balance",
+    "neutral_mass": 304.207146,
+    "ion_mz": 305.21442247
+  },
+  "iTRAQ113": {
+    "label_type": "iTRAQ",
+    "molecule_type": "reporter",
+    "ion_mz": 113.1078
+  },
+  "iTRAQ114": {
+    "label_type": "iTRAQ",
+    "molecule_type": "reporter",
+    "ion_mz": 114.1112
+  },
+  "iTRAQ115": {
+    "label_type": "iTRAQ",
+    "molecule_type": "reporter",
+    "ion_mz": 115.1082
+  },
+  "iTRAQ116": {
+    "label_type": "iTRAQ",
+    "molecule_type": "reporter",
+    "ion_mz": 116.1116
+  },
+  "iTRAQ117": {
+    "label_type": "iTRAQ",
+    "molecule_type": "reporter",
+    "ion_mz": 117.1149
+  },
+  "iTRAQ118": {
+    "label_type": "iTRAQ",
+    "molecule_type": "reporter",
+    "ion_mz": 118.112
+  },
+  "iTRAQ119": {
+    "label_type": "iTRAQ",
+    "molecule_type": "reporter",
+    "ion_mz": 119.1153
+  },
+  "iTRAQ121": {
+    "label_type": "iTRAQ",
+    "molecule_type": "reporter",
+    "ion_mz": 121.122
+  },
+  "iTRAQ4plex": {
+    "label_type": "iTRAQ4plex",
+    "molecule_type": "reporter+balance",
+    "neutral_mass": 144.102063,
+    "ion_mz": 145.10933947
+  },
+  "iTRAQ8plex": {
+    "label_type": "iTRAQ8plex",
+    "molecule_type": "reporter+balance",
+    "neutral_mass": 304.205360,
+    "ion_mz": 305.21263647
+  },
+  "TMT126-ETD": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 114.127725
+  },
+  "TMT127N-ETD": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 115.12476
+  },
+  "TMT127C-ETD": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 114.127725
+  },
+  "TMT128N-ETD": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 115.12476
+  },
+  "TMT128C-ETD": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 116.134433
+  },
+  "TMT129N-ETD": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 117.131468
+  },
+  "TMT129C-ETD": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 116.134433
+  },
+  "TMT130N-ETD": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 117.131468
+  },
+  "TMT130C-ETD": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 118.141141
+  },
+  "TMT131N-ETD": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 119.138176
+  },
+  "TMT131C-ETD": {
+    "label_type": "TMT",
+    "molecule_type": "reporter",
+    "ion_mz": 118.141141
+  }
+}

implementations/python/mzpaf/reference.py

@@ -1,8 +1,8 @@
-
 import json
+from importlib.resources import open_text
 
 from dataclasses import dataclass, field, asdict
-from typing import Any, List, Optional, Pattern, Dict, Tuple, Type, Union
+from typing import Any, List, Optional, Pattern, Dict, Tuple, Type, Union, ClassVar
 
 PROTON = 1.00727646677
 
@@ -14,26 +14,31 @@ def _mass_charge_ratio(neutral_mass, z, charge_carrier=PROTON):
     return (neutral_mass + (z * charge_carrier)) / abs(z)
 
 
-@dataclass
+@dataclass(frozen=True)
 class ReferenceMolecule:
     name: str
     molecule_type: str
-    label_type: str = field(default=None)
+    cv_term: Optional[str] = field(default=None)
+    label_type: Optional[str] = field(default=None)
     neutral_mass: float = field(default=None)
     ion_mz: float = field(default=None)
-    chemical_formula: str = field(default=None)
-    ion_chemical_formula: str = field(default=None)
+    chemical_formula: Optional[str] = field(default=None)
+    ion_chemical_formula: Optional[str] = field(default=None)
     references: List[str] = field(default_factory=list)
 
+    _registry: ClassVar[Dict[str, 'ReferenceMolecule']] = None
+
     def __post_init__(self):
         if self.neutral_mass is None:
             if self.ion_mz is not None:
-                self.neutral_mass = _neutral_mass(self.ion_mz, 1)
+                object.__setattr__(self, "neutral_mass",
+                                   _neutral_mass(self.ion_mz, 1))
             else:
                 raise ValueError("Must provide at least one of `neutral_mass` or `ion_mz`!")
         elif self.ion_mz is None:
             if self.neutral_mass is not None:
-                self.ion_mz = _mass_charge_ratio(self.neutral_mass, 1)
+                object.__setattr__(self, 'neutral_mass',
+                                   _mass_charge_ratio(self.neutral_mass, 1))
             else:
                 raise ValueError("Must provide at least one of `neutral_mass` or `ion_mz`!")
 
@@ -44,6 +49,22 @@ def to_dict(self):
     def from_dict(cls, state, **kwargs):
         return cls(**state)
 
+    @classmethod
+    def _load_registry(cls):
+        with open_text("mzpaf.data", "reference_molecules.json") as stream:
+            data = json.load(stream)
+        cls._registry = {}
+        for k, v in data.items():
+            v['name'] = k
+            cls._registry[k] = cls.from_dict(v)
+
+    @classmethod
+    def get(cls, name: str) -> 'ReferenceMolecule':
+        if cls._registry is None:
+            cls._load_registry()
+        return cls._registry[name]
+
+
 
 def load_json(stream) -> Dict[str, ReferenceMolecule]:
     if isinstance(stream, dict):

specification/reference_data/reference_molecule_schema.json

@@ -17,6 +17,12 @@
                     "type": "string",
                     "description": "The formal name for this molecule by which it should be referenced"
                 },
+                "cv_term": {
+                    "type": "array",
+                    "items": {
+                        "type":"string"
+                    }
+                },
                 "neutral_mass": {
                     "type": "number",
                     "description": "The neutral mass of the molecule not including any charge or charge carrier"