use sip int id

Rappsilber-Laboratory · Jul 28, 2024 · 07d6373 · 07d6373
1 parent 8da471a
commit 07d6373
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 12 deletions.
diff --git a/models/match.py b/models/match.py
@@ -21,12 +21,12 @@ class Match(Base):
     scores: Mapped[Optional[dict[str, Any]]] = mapped_column(JSON, nullable=True)
     exp_mz: Mapped[float] = mapped_column(FLOAT, nullable=True)
     calc_mz: Mapped[float] = mapped_column(FLOAT, nullable=True)
-    sil_id: Mapped[str] = mapped_column(Text, nullable=True)  # null if from csv file
+    sip_id: Mapped[int] = mapped_column(Integer, nullable=True)  # null if from csv file
     __table_args__ = (
         ForeignKeyConstraint(
-            ["sil_id", "upload_id"],
-            ["analysiscollectionspectrumidentification.spectrum_identification_list_ref",
-             "analysiscollectionspectrumidentification.upload_id"],
+            ["sip_id", "upload_id"],
+            ["spectrumidentificationprotocol.id",
+             "spectrumidentificationprotocol.upload_id"],
         ),
         ForeignKeyConstraint(
             ["pep1_id", "upload_id"],

diff --git a/parser/MzIdParser.py b/parser/MzIdParser.py
@@ -44,6 +44,8 @@ def __init__(self, mzid_path, temp_dir, peak_list_dir, writer, logger):
 
         self.peak_list_readers = {}  # peak list readers indexed by spectraData_ref
         self.spectra_data_id_lookup = {}  # spectra_data_ref to spectra_data_id lookup
+        self.sip_ref_to_sip_id_lookup = {}  # sip_ref to sip_id lookup
+        self.sil_ref_to_protocol_id_lookup = {}  # sip_ref to sip_id lookup
         self.temp_dir = temp_dir
         if not self.temp_dir.endswith('/'):
             self.temp_dir += '/'
@@ -385,6 +387,7 @@ def parse_analysis_protocol_collection(self):
                     })
 
             sid_protocols.append(data)
+            self.sip_ref_to_sip_id_lookup[sid_protocol['id']] = sip_int_id
             sip_int_id += 1
 
         self.mzid_reader.reset()
@@ -421,7 +424,7 @@ def parse_analysis_collection(self):
                 'search_database_refs': search_database_refs
             }
             spectrum_identification.append(si_data)
-
+            self.sil_ref_to_protocol_id_lookup[si['spectrumIdentificationList_ref']] = self.sip_ref_to_sip_id_lookup[si['spectrumIdentificationProtocol_ref']]
         self.mzid_reader.reset()
         self.logger.info('parsing AnalysisCollection - done. Time: {} sec'.format(
             round(time() - start_time, 2)))
@@ -743,7 +746,7 @@ def main_loop(self):
                             'scores': psm_level_stats,
                             'exp_mz': spec_id_item['experimentalMassToCharge'],
                             'calc_mz': calculated_mass_to_charge,
-                            'sil_id': sil_id,
+                            'sip_id': self.sil_ref_to_protocol_id_lookup[sil_id],
                             'multiple_spectra_identification_id': msi_id,
                             'multiple_spectra_identification_pc': msi_pc,
                         }

diff --git a/tests/test_mzid_1_3_0.py b/tests/test_mzid_1_3_0.py
@@ -31,42 +31,42 @@ def test_psql_multi_spectra_mzid_parser(tmpdir, db_info, use_database, engine):
         assert results[0].pep2_id == 'p2'
         assert results[0].multiple_spectra_identification_id == 1234
         assert results[0].multiple_spectra_identification_pc == 'P'
-        assert results[0].sil_id == 'sil_HCD'
+        assert results[0].sip_id == 0 #  'sil_HCD'
 
         assert results[1].id == 'ETD_SII_0'
         assert results[1].pep1_id == 'p1'
         assert results[1].pep2_id == 'p2'
         assert results[1].multiple_spectra_identification_id == 1234
         assert results[1].multiple_spectra_identification_pc == 'P'
-        assert results[1].sil_id == 'sil_ETD'
+        assert results[1].sip_id == 1 #  'sil_ETD'
 
         assert results[2].id == 'MS3_SII_0'
         assert results[2].pep1_id == 'p1_a'
         assert results[2].pep2_id is None
         assert results[2].multiple_spectra_identification_id == 1234
         assert results[2].multiple_spectra_identification_pc == 'C'
-        assert results[2].sil_id == 'sil_MS3'
+        assert results[2].sip_id == 2 # 'sil_MS3'
 
         assert results[3].id == 'MS3_SII_1'
         assert results[3].pep1_id == 'p2_t'
         assert results[3].pep2_id is None
         assert results[3].multiple_spectra_identification_id == 1234
         assert results[3].multiple_spectra_identification_pc == 'C'
-        assert results[3].sil_id == 'sil_MS3'
+        assert results[3].sip_id == 2 # 'sil_MS3'
 
         assert results[4].id == 'MS3_SII_2'
         assert results[4].pep1_id == 'p1_t'
         assert results[4].pep2_id is None
         assert results[4].multiple_spectra_identification_id == 1234
         assert results[4].multiple_spectra_identification_pc == 'C'
-        assert results[4].sil_id == 'sil_MS3'
+        assert results[4].sip_id == 2 # 'sil_MS3'
 
         assert results[5].id == 'MS3_SII_3'
         assert results[5].pep1_id == 'p2_a'
         assert results[5].pep2_id is None
         assert results[5].multiple_spectra_identification_id == 1234
         assert results[5].multiple_spectra_identification_pc == 'C'
-        assert results[5].sil_id == 'sil_MS3'
+        assert results[5].sip_id == 2 # 'sil_MS3'
 
     engine.dispose()