Edges from complexes are now represented as edges from the individual…

… components of the complex, with a 'comple_context' edge property to record the complex name the edge was derived from.
RobokopU24 · Oct 3, 2023 · 7d15da1 · 7d15da1
1 parent 08aa891
commit 7d15da1
Show file tree

Hide file tree

Showing 3 changed files with 114 additions and 35 deletions.
diff --git a/parsers/Reactome/src/loadReactome.py b/parsers/Reactome/src/loadReactome.py
@@ -71,10 +71,12 @@
 # GenomeEncodedEntity: EntityWITHACCESSIONEDSEQUENCE(Protein), EntityWITHACCESSIONEDSEQUENCE(Gene and transcript), EntityWITHACCESSIONEDSEQUENCE(DNA), EntityWITHACCESSIONEDSEQUENCE(RNA)
 CROSS_MAPPING = ('EntityWithAccessionedSequence','GenomeEncodedEntity', 'SimpleEntity', 'Drug', 'Complex', 'Polymer')
 
-TO_WRITE = ('Provenance/Include', 'Attribute/Include') #Descriptive features of other existing nodes eg Summation
+TO_WRITE = ('Provenance/Include','Attribute/Include') #Descriptive features of other existing nodes eg Summation
 TO_MAP = ('IDMapping/Include', ) # Maps the external identifier of a node from another node
 TO_INCLUDE = ('Include',)
 RDF_EDGES_TO_INCLUDE = ('RDF_edges/Include',)
+MOLE_COMPLEX = ('Include/Complex',) # 
+TO_SWITCH_MOLE_COMPLEX = ('Include/SwitchSO/Complex',) # 
 TO_SWITCH = ('Include/SwitchSO', )
 
 ##############
@@ -173,6 +175,18 @@ def extract_data(self, neo4j_driver) -> dict:
                         f"AND any(x in labels(b) WHERE x in ['Complex','GenomeEncodedEntity','EntityWithAccessionedSequence']) " \
                         f"RETURN a, labels(a) as a_labels, id(a) as a_id, type(r) as r_type, b, labels(b) as b_labels, id(b) as b_id, labels(d) as regulationType"
                 queries_to_include.append(cypher_query)
+            elif line[INCLUDE_COLUMN] in TO_SWITCH_MOLE_COMPLEX:
+                cypher_query = f"MATCH (b)<-[r:hasComponent]-(c:{line[SUBJECT_COLUMN]})-[r1:{line[PREDICATE_COLUMN]}]->(a:{line[OBJECT_COLUMN]}) " \
+                        f"RETURN a, labels(a) as a_labels, id(a) as a_id, type(r) as r_type, b, labels(b) as b_labels, id(b) as b_id, c.name as complex_context"
+                # Remove if map_ids is not needed.
+                #cypher_query = self.map_ids(line[SUBJECT_COLUMN], line[PREDICATE_COLUMN], line[OBJECT_COLUMN], switch=True)
+                queries_to_include.append(cypher_query)
+            elif line[INCLUDE_COLUMN] in MOLE_COMPLEX:
+                cypher_query = f"MATCH (a)<-[r:hasComponent]-(c:{line[SUBJECT_COLUMN]})-[r1:{line[PREDICATE_COLUMN]}]->(b:{line[OBJECT_COLUMN]}) " \
+                        f"RETURN a, labels(a) as a_labels, id(a) as a_id, type(r) as r_type, b, labels(b) as b_labels, id(b) as b_id, c.name as complex_context"
+                # Remove if map_ids is not needed.
+                #cypher_query = self.map_ids(line[SUBJECT_COLUMN], line[PREDICATE_COLUMN], line[OBJECT_COLUMN], switch=True)
+                queries_to_include.append(cypher_query)
             elif line[INCLUDE_COLUMN] in TO_SWITCH:
                 cypher_query = f"MATCH (b:{line[SUBJECT_COLUMN]})-[r:{line[PREDICATE_COLUMN]}]->(a:{line[OBJECT_COLUMN]}) " \
                         f"RETURN a, labels(a) as a_labels, id(a) as a_id, type(r) as r_type, b, labels(b) as b_labels, id(b) as b_id"
@@ -238,7 +252,7 @@ def extract_data(self, neo4j_driver) -> dict:
     def get_reference_entity_mapping(self, neo4j_session):
         reference_entity_mapping = {}
         # The following line excludes Pathways from ID mapping because we only want to map them to GO terms, like 2 lines below.
-        reference_entity_query = "MATCH (a)-[:referenceEntity|crossReference]->(b) WHERE NOT('Pathway' in labels(a)) return id(a) as identity, b as reference, labels(b) as ref_labels"
+        reference_entity_query = "MATCH (a)-[r:referenceEntity|crossReference]->(b) WHERE NOT('Pathway' in labels(a)) return id(a) as identity, b as reference, labels(b) as ref_labels"
         goBioProcess_query = "MATCH (a:Pathway)-[r:goBiologicalProcess]->(b:GO_Term) WHERE replace(toLower(a.displayName),'-',' ') = replace(toLower(b.displayName),'-',' ') return id(a) as identity, b as reference, labels(b) as ref_labels"
         reference_entity_result = neo4j_session.run(reference_entity_query)
         goBioProcess_query = neo4j_session.run(goBioProcess_query)
@@ -289,11 +303,20 @@ def write_neo4j_result_to_file(self, result: neo4j.Result, reference_entity_mapp
             if node_a_id and node_b_id:
                 if "regulationType" in record_data.keys():
                     if any("positive" in x.lower() for x in record_data['regulationType']):
-                        self.process_edge_from_neo4j(node_a_id, record_data['r_type'], node_b_id, regulationType='positive')
+                        if "complex_context" in record_data.keys():
+                            self.process_edge_from_neo4j(node_a_id, record_data['r_type'], node_b_id, regulationType='positive', complex_context=record_data['complex_context'])
+                        else:
+                            self.process_edge_from_neo4j(node_a_id, record_data['r_type'], node_b_id, regulationType='positive')
                     elif any("negative" in x.lower() for x in record_data['regulationType']):
-                        self.process_edge_from_neo4j(node_a_id, record_data['r_type'], node_b_id, regulationType='negative')
+                        if "complex_context" in record_data.keys():
+                            self.process_edge_from_neo4j(node_a_id, record_data['r_type'], node_b_id, regulationType='negative', complex_context=record_data['complex_context'])
+                        else:
+                            self.process_edge_from_neo4j(node_a_id, record_data['r_type'], node_b_id, regulationType='negative')
                 else:
-                    self.process_edge_from_neo4j(node_a_id, record_data['r_type'], node_b_id, regulationType=None)
+                    if "complex_context" in record_data.keys():
+                        self.process_edge_from_neo4j(node_a_id, record_data['r_type'], node_b_id, regulationType=None, complex_context=record_data['complex_context'])
+                    else:
+                        self.process_edge_from_neo4j(node_a_id, record_data['r_type'], node_b_id, regulationType=None)
                 record_count += 1
             else:
                 skipped_record_count += 1
@@ -380,32 +403,55 @@ def process_node_from_neo4j(self, reference_entity_mapping, node_identity, node:
         self.dbid_to_node_id_lookup[node['dbId']] = node_id
         """
 
-    def process_edge_from_neo4j(self, subject_id: str, relationship_type: str, object_id: str, regulationType=None):
+    def process_edge_from_neo4j(self, subject_id: str, relationship_type: str, object_id: str, regulationType=None, complex_context=None):
         predicate = PREDICATE_MAPPING.get(relationship_type, None)
         if predicate:
             if regulationType == None:
-                output_edge = kgxedge(
-                    subject_id=subject_id,
-                    object_id=object_id,
-                    predicate=predicate,
-                    primary_knowledge_source=self.provenance_id
-                )
+                if complex_context != None:
+                    output_edge = kgxedge(
+                        subject_id=subject_id,
+                        object_id=object_id,
+                        predicate=predicate,
+                        edgeprops = {'complex_context':complex_context},
+                        primary_knowledge_source=self.provenance_id
+                    )
+                else:
+                    output_edge = kgxedge(
+                        subject_id=subject_id,
+                        object_id=object_id,
+                        predicate=predicate,
+                        primary_knowledge_source=self.provenance_id
+                    )
             else:
                 if regulationType == "positive":
                     direction = 'increases'
                 elif regulationType == "negative":
                     direction = 'decreases'
-                output_edge = kgxedge(
-                    subject_id=subject_id,
-                    object_id=object_id,
-                    predicate=predicate,
-                    edgeprops={
-                        'qualified_predicate':'biolink:causes',
-                        'object_direction_qualifier':direction,
-                        'object_aspect_qualifier':'expression'
-                    },
-                    primary_knowledge_source=self.provenance_id
-                )
+                if complex_context != None:
+                    output_edge = kgxedge(
+                        subject_id=subject_id,
+                        object_id=object_id,
+                        predicate=predicate,
+                        edgeprops={
+                            'qualified_predicate':'biolink:causes',
+                            'object_direction_qualifier':direction,
+                            'object_aspect_qualifier':'expression',
+                            'complex_context':complex_context
+                        },
+                        primary_knowledge_source=self.provenance_id
+                    )
+                else:
+                    output_edge = kgxedge(
+                        subject_id=subject_id,
+                        object_id=object_id,
+                        predicate=predicate,
+                        edgeprops={
+                            'qualified_predicate':'biolink:causes',
+                            'object_direction_qualifier':direction,
+                            'object_aspect_qualifier':'expression',
+                        },
+                        primary_knowledge_source=self.provenance_id
+                    )
             self.output_file_writer.write_kgx_edge(output_edge)
         else:
             self.logger.warning(f'A predicate could not be mapped for relationship type {relationship_type}')

diff --git a/parsers/Reactome/src/reactomeContents_CriticalTriples.csv b/parsers/Reactome/src/reactomeContents_CriticalTriples.csv
@@ -23,18 +23,10 @@ EntityWithAccessionedSequence,cellType,CellType,Include,
 Drug,cellType,CellType,Include,
 Pathway,disease,Disease,Include/SwitchSO,
 ReactionLikeEvent,disease,Disease,Include/SwitchSO,
-Complex,hasComponent,EntityWithAccessionedSequence,Include,
-Complex,species,Species,Include,
-Complex,compartment,GO_Term,Include,
-Complex,hasComponent,Complex,Include,
-ReactionLikeEvent,output,Complex,Include,
-ReactionLikeEvent,input,Complex,Include,
-Complex,hasComponent,SimpleEntity,Include,
-Complex,hasComponent,GenomeEncodedEntity,Include,
-Complex,includedLocation,GO_Term,Include,
-Complex,disease,Disease,Include/SwitchSO,
-Complex,hasComponent,Polymer,Include,
-Regulation,regulator,Complex,RDF_edges/Include,
+ReactionLikeEvent,output,Complex,Include/Complex,
+ReactionLikeEvent,input,Complex,Include/Complex,
+Complex,disease,Disease,Include/SwitchSO/Complex,
+Regulation,regulator,Complex,Include/Complex,
 Regulation,regulator,EntityWithAccessionedSequence,RDF_edges/Include,
 Regulation,regulator,SimpleEntity,RDF_edges/Include,
 Regulation,regulator,Drug,RDF_edges/Include,

diff --git a/parsers/Reactome/src/reactomeContents_CriticalTriples_withComplexes.csv b/parsers/Reactome/src/reactomeContents_CriticalTriples_withComplexes.csv
@@ -0,0 +1,41 @@
+Subject,Predicate,Object,Decision,Notes:
+Pathway,hasEvent,ReactionLikeEvent,Include,
+ReactionLikeEvent,input,SimpleEntity,Include,
+ReactionLikeEvent,output,SimpleEntity,Include,
+ReactionLikeEvent,precedingEvent,ReactionLikeEvent,Include,
+ReactionLikeEvent,input,EntityWithAccessionedSequence,Include,
+ReactionLikeEvent,output,EntityWithAccessionedSequence,Include,
+ReactionLikeEvent,input,GenomeEncodedEntity,Include,
+ReactionLikeEvent,output,GenomeEncodedEntity,Include,
+Pathway,hasEvent,Pathway,Include,
+ReactionLikeEvent,compartment,GO_Term,Include,
+Pathway,compartment,GO_Term,Include,
+Pathway,goBiologicalProcess,GO_BiologicalProcess,Include,
+ReactionLikeEvent,goBiologicalProcess,GO_BiologicalProcess,Include,
+ReactionLikeEvent,precedingEvent,ReactionLikeEvent,Include,
+Pathway,precedingEvent,Pathway,Include,
+ReactionLikeEvent,input,Polymer,Include,
+ReactionLikeEvent,output,Polymer,Include,
+ReactionLikeEvent,input,Drug,Include,
+ReactionLikeEvent,output,Drug,Include,
+SimpleEntity,cellType,CellType,Include,
+EntityWithAccessionedSequence,cellType,CellType,Include,
+Drug,cellType,CellType,Include,
+Pathway,disease,Disease,Include/SwitchSO,
+ReactionLikeEvent,disease,Disease,Include/SwitchSO,
+Complex,hasComponent,EntityWithAccessionedSequence,Include,
+Complex,species,Species,Include,
+Complex,compartment,GO_Term,Include,
+Complex,hasComponent,Complex,Include,
+ReactionLikeEvent,output,Complex,Include,
+ReactionLikeEvent,input,Complex,Include,
+Complex,hasComponent,SimpleEntity,Include,
+Complex,hasComponent,GenomeEncodedEntity,Include,
+Complex,includedLocation,GO_Term,Include,
+Complex,disease,Disease,Include/SwitchSO,
+Complex,hasComponent,Polymer,Include,
+Regulation,regulator,Complex,RDF_edges/Include,
+Regulation,regulator,EntityWithAccessionedSequence,RDF_edges/Include,
+Regulation,regulator,SimpleEntity,RDF_edges/Include,
+Regulation,regulator,Drug,RDF_edges/Include,
+Regulation,regulator,GenomeEncodedEntity,RDF_edges/Include,