Merge pull request #5 from xgaia/4.1.0

Release 4.1.0

README.md

@@ -72,6 +72,13 @@ Example with [NeXtProt](https://sparql.nextprot.org):
 abstractor -s https://sparql.nextprot.org -o nextprot_abstraction.ttl
 ```
 
+
+Example with [Uniprot](https://sparql.uniprot.org), using OWL:
+
+```bash
+abstractor -s https://sparql.nextprot.org -o nextprot_abstraction.ttl -m owl
+```
+
 #### With a RDF file
 
 ```bash

abstractor

@@ -24,20 +24,22 @@ class Abstractor(object):
 
         parser.add_argument("-o", "--output", type=str, help="Output file", default="abstraction.rdf")
         parser.add_argument("-f", "--output-format", choices=['xml', 'turtle', 'nt'], help="RDF format", default="turtle")
-        parser.add_argument("--owl", default=False, action='store_true', help="Use OWL ontology")
+
+        parser.add_argument("-m", "--mode", choices=["all", "batch", "owl", "askomics"], help="Scan mode: all: 3 queries to get all entities,\
+         relation and attributes. batch: 3 queries for each entity. owl: 3 queries using existant owl ontology. askomics: 3 queries using askomics ontology (NOT IMPLEMENTED YET)", default="all")
 
         parser.add_argument("-v", "--verbosity", action="count", help="increase output verbosity")
 
         self.args = parser.parse_args()
 
         logging_level = logging.CRITICAL
-        if self.args.verbosity == 1:
+        if self.args.verbosity is None or self.args.verbosity == 1:
             logging_level = logging.ERROR
-        if self.args.verbosity == 2:
+        elif self.args.verbosity == 2:
             logging_level = logging.WARNING
-        if self.args.verbosity == 3:
+        elif self.args.verbosity == 3:
             logging_level = logging.INFO
-        if self.args.verbosity > 3:
+        elif self.args.verbosity > 3:
             logging_level = logging.DEBUG
 
         logging.basicConfig(level=logging_level)
@@ -52,9 +54,36 @@ class Abstractor(object):
         if self.args.source_type == "sparql":
             rdf.add_location(self.args.source)
 
-        # Use owl ontology
-        if self.args.owl:
-            logging.debug("Use OWL Ontology")
+        if self.args.mode == "all":
+            logging.debug("Get entities and relation")
+            rdf.add_entities_and_relations(sparql.process_query(library.entities_and_relations))
+            logging.debug("Get decimal attributes")
+            rdf.add_decimal_attributes(sparql.process_query(library.entities_and_numeric_attributes))
+            logging.debug("Get text attributes")
+            rdf.add_text_attributes(sparql.process_query(library.entities_and_text_attributes))
+
+        elif self.args.mode == "batch":
+            logging.debug("Get all entities, then, get relations and attributes for each entity")
+            entities = sparql.process_query(library.get_entities)
+            rdf.add_entities(entities)
+            for entity_dict in entities:
+                entity = entity_dict["entity"]
+                if rdf.check_entity(entity):
+                    # relation
+                    relations = sparql.process_query(library.get_relation_for_entity(entity))
+                    for relation_dict in relations:
+                        rdf.add_relation(entity, relation_dict["relation"], relation_dict["target_entity"])
+                    # numeric attribute
+                    attributes = sparql.process_query(library.get_numeric_attribute_for_entity(entity))
+                    for attribute_dict in attributes:
+                        rdf.add_attribute(entity, attribute_dict["attribute"])
+
+                    text_attributes = sparql.process_query(library.get_text_attribute_for_entity(entity))
+                    for attribute_dict in text_attributes:
+                        rdf.add_attribute(entity, attribute_dict["attribute"], decimal=False)
+
+        elif self.args.mode == "owl":
+            logging.debug("Use OWL ontology")
             result = sparql.process_query(library.ontologies)
             for res in result:
                 logging.debug(res["ontology"])
@@ -65,14 +94,9 @@ class Abstractor(object):
                 logging.debug("Get text attributes")
                 rdf.add_text_attributes(sparql.process_query(library.entities_and_text_attributes_with_ontology(res["ontology"])))
 
-        # All relations
-        else:
-            logging.debug("Get entities and relation")
-            rdf.add_entities_and_relations(sparql.process_query(library.entities_and_relations))
-            logging.debug("Get decimal attributes")
-            rdf.add_decimal_attributes(sparql.process_query(library.entities_and_numeric_attributes))
-            logging.debug("Get text attributes")
-            rdf.add_text_attributes(sparql.process_query(library.entities_and_text_attributes))
+        elif self.args.mode == "askomics":
+            logging.debug("Use AskOmics ontology")
+            raise NotImplementedError
 
         logging.debug("Write RDF ({}) into {}".format(self.args.output_format, self.args.output))
         rdf.graph.serialize(destination=self.args.output, format=self.args.output_format, encoding="utf-8" if self.args.output_format == "turtle" else None)

libabstractor/QueryLibrary.py

@@ -8,6 +8,92 @@ def __init__(self):
         """init"""
         pass
 
+    @property
+    def get_entities(self):
+        """Sparql query to get all entities
+
+        Returns
+        -------
+        str
+            SPARQL query
+        """
+        return textwrap.dedent('''
+        SELECT DISTINCT ?entity
+        WHERE {
+            ?instance a ?entity .
+        }
+        ''')
+
+    def get_relation_for_entity(self, entity):
+        """Sparql query to get all relations of an entity
+
+        Parameters
+        ----------
+        entity : string
+            The source entity
+
+        Returns
+        -------
+        str
+            SPARQL query
+        """
+        return textwrap.dedent('''
+        SELECT DISTINCT ?relation ?target_entity
+        WHERE {{
+            <{}> ?relation ?target_entity .
+            ?instance_of_target a ?target_entity .
+        }}
+        '''.format(entity))
+
+    def get_numeric_attribute_for_entity(self, entity):
+        """Sparql query to get all attribute of an entity
+
+        Parameters
+        ----------
+        entity : string
+            The source entity
+
+        Returns
+        -------
+        str
+            SPARQL query
+        """
+        return textwrap.dedent('''
+        SELECT DISTINCT ?attribute
+        WHERE {{
+            # Get entities
+            ?instance_of_entity a <{}> .
+            # Attributes
+            ?instance_of_entity ?attribute ?value .
+            FILTER (isNumeric(?value))
+        }}
+        '''.format(entity))
+
+    def get_text_attribute_for_entity(self, entity):
+        """Sparql query to get all text attribute of an entity
+
+        Parameters
+        ----------
+        entity : string
+            The source entity
+
+        Returns
+        -------
+        str
+            SPARQL query
+        """
+        return textwrap.dedent('''
+        SELECT DISTINCT ?attribute
+        WHERE {{
+            # Get entities
+            ?instance_of_entity a <{}> .
+            # Attributes
+            ?instance_of_entity ?attribute ?value .
+            FILTER (isLiteral(?value))
+            FILTER (!isNumeric(?value))
+        }}
+        '''.format(entity))
+
     @property
     def entities_and_relations(self):
         """Sparql query to get entities and relations
@@ -18,13 +104,20 @@ def entities_and_relations(self):
             SPARQL query
         """
         return textwrap.dedent('''
-        SELECT DISTINCT ?source_entity ?relation ?target_entity
+        SELECT DISTINCT ?source_entity ?relation ?target_entity ?mother_source ?mother_target
         WHERE {
             # Get entities
             ?instance_of_source a ?source_entity .
             ?instance_of_target a ?target_entity .
             # Relations
             ?instance_of_source ?relation ?instance_of_target .
+
+            OPTIONAL {{
+                ?source_entity rdfs:subClassOf ?mother_source .
+            }}
+            OPTIONAL {{
+                ?target_entity rdfs:subClassOf ?mother_target .
+            }}
         }
         ''')
 
@@ -101,7 +194,7 @@ def entities_and_relations_with_ontology(ontology):
             SPARQL query
         """
         return textwrap.dedent('''
-        SELECT DISTINCT ?source_entity ?relation ?target_entity
+        SELECT DISTINCT ?source_entity ?relation ?target_entity ?mother_source ?mother_target
         WHERE {{
             ?source_entity a owl:Class .
             ?source_entity rdfs:isDefinedBy <{ontology}> .
@@ -110,8 +203,18 @@ def entities_and_relations_with_ontology(ontology):
             ?target_entity rdfs:isDefinedBy <{ontology}> .
 
             ?relation a owl:ObjectProperty .
-            ?relation rdfs:domain ?source_entity .
             ?relation rdfs:range ?target_entity .
+            {{
+                ?relation rdfs:domain/(owl:unionOf/(rdf:rest*)/rdf:first) ?source_entity .
+            }} UNION {{
+                ?relation rdfs:domain ?source_entity .
+            }}
+            OPTIONAL {{
+                ?source_entity rdfs:subClassOf ?mother_source .
+            }}
+            OPTIONAL {{
+                ?target_entity rdfs:subClassOf ?mother_target .
+            }}
         }}
         '''.format(ontology=ontology))
 
@@ -132,9 +235,13 @@ def entities_and_numeric_attributes_with_ontology(ontology):
             ?entity rdfs:isDefinedBy <{ontology}> .
             # Attribute
             ?attribute a owl:DatatypeProperty .
-            ?attribute rdfs:domain ?entity .
             ?attribute rdfs:range ?range .
             VALUES ?range {{ xsd:float xsd:int }} .
+            {{
+                ?attribute rdfs:domain/(owl:unionOf/(rdf:rest*)/rdf:first) ?entity .
+            }} UNION {{
+                ?attribute rdfs:domain ?entity .
+            }}
         }}
         '''.format(ontology=ontology))
 
@@ -155,8 +262,12 @@ def entities_and_text_attributes_with_ontology(ontology):
             ?entity rdfs:isDefinedBy <{ontology}> .
             # Attribute
             ?attribute a owl:DatatypeProperty .
-            ?attribute rdfs:domain ?entity .
             ?attribute rdfs:range ?range .
             VALUES ?range {{ xsd:string }} .
+            {{
+                ?attribute rdfs:domain/(owl:unionOf/(rdf:rest*)/rdf:first) ?entity .
+            }} UNION {{
+                ?attribute rdfs:domain ?entity .
+            }}
         }}
         '''.format(ontology=ontology))

libabstractor/RdfGraph.py

@@ -59,6 +59,58 @@ def add_location(self, location):
         """
         self.graph.add((rdflib.BNode("graph"), self.prov.atLocation, rdflib.Literal(location)))
 
+    def add_entities(self, sparql_result):
+        """Add entities
+
+        Parameters
+        ----------
+        sparql_result : list
+            Sparql result
+        """
+        for result in sparql_result:
+            if self.check_entity(result["entity"]):
+                self.graph.add((rdflib.URIRef(result["entity"]), rdflib.RDF.type, self.namespace_internal["entity"]))
+                self.graph.add((rdflib.URIRef(result["entity"]), rdflib.RDF.type, self.namespace_internal["startPoint"]))
+                self.graph.add((rdflib.URIRef(result["entity"]), rdflib.RDF.type, rdflib.OWL.Class))
+                self.graph.add((rdflib.URIRef(result["entity"]), self.namespace_internal["instancesHaveNoLabels"], rdflib.Literal(True)))
+                self.graph.add((rdflib.URIRef(result["entity"]), rdflib.RDFS.label, rdflib.Literal(self.get_label(result["entity"]))))
+
+    def add_relation(self, source_entity, relation, target_entity):
+        """Add a relation
+
+        Parameters
+        ----------
+        source_entity : str
+            Source URI
+        relation : str
+            Relation URI
+        target_entity : str
+            Target URI
+        """
+        # Relation
+        if self.check_entity(relation):
+            self.graph.add((rdflib.URIRef(relation), rdflib.RDF.type, rdflib.OWL.ObjectProperty))
+            self.graph.add((rdflib.URIRef(relation), rdflib.RDF.type, self.namespace_internal["AskomicsRelation"]))
+            self.graph.add((rdflib.URIRef(relation), rdflib.RDFS.label, rdflib.Literal(self.get_label(relation))))
+            self.graph.add((rdflib.URIRef(relation), rdflib.RDFS.domain, rdflib.URIRef(source_entity)))
+            self.graph.add((rdflib.URIRef(relation), rdflib.RDFS.range, rdflib.URIRef(target_entity)))
+
+    def add_attribute(self, entity, attribute, decimal=True):
+        """Add attribute
+
+        Parameters
+        ----------
+        entity : str
+            Source URI
+        attribute : str
+            Attribue URI
+        """
+        if self.check_entity(entity):
+            self.graph.add((rdflib.URIRef(attribute), rdflib.RDF.type, rdflib.OWL.DatatypeProperty))
+            self.graph.add((rdflib.URIRef(attribute), rdflib.RDFS.label, rdflib.Literal(self.get_label(attribute))))
+            self.graph.add((rdflib.URIRef(attribute), rdflib.RDFS.domain, rdflib.URIRef(entity)))
+            self.graph.add((rdflib.URIRef(attribute), rdflib.RDFS.range, rdflib.XSD.decimal if decimal else rdflib.XSD.string))
+
     def add_entities_and_relations(self, sparql_result):
         """Add entities and relation in the rdf graph
 
@@ -74,6 +126,8 @@ def add_entities_and_relations(self, sparql_result):
             source_entity = result["source_entity"]
             target_entity = result["target_entity"]
             relation = result["relation"]
+            mother_source = result["mother_source"] if "mother_source" in result else None
+            mother_target = result["mother_target"] if "mother_target" in result else None
 
             # Source entity
             if self.check_entity(source_entity) and source_entity not in entities:
@@ -83,6 +137,8 @@ def add_entities_and_relations(self, sparql_result):
                 self.graph.add((rdflib.URIRef(source_entity), rdflib.RDF.type, rdflib.OWL.Class))
                 self.graph.add((rdflib.URIRef(source_entity), self.namespace_internal["instancesHaveNoLabels"], rdflib.Literal(True)))
                 self.graph.add((rdflib.URIRef(source_entity), rdflib.RDFS.label, rdflib.Literal(self.get_label(source_entity))))
+                if mother_source:
+                    self.graph.add((rdflib.URIRef(source_entity), rdflib.RDFS.subClassOf, rdflib.URIRef(mother_source)))
 
             # Target entity
             if self.check_entity(target_entity) and target_entity not in entities:
@@ -92,6 +148,8 @@ def add_entities_and_relations(self, sparql_result):
                 self.graph.add((rdflib.URIRef(target_entity), rdflib.RDF.type, rdflib.OWL.Class))
                 self.graph.add((rdflib.URIRef(target_entity), self.namespace_internal["instancesHaveNoLabels"], rdflib.Literal(True)))
                 self.graph.add((rdflib.URIRef(target_entity), rdflib.RDFS.label, rdflib.Literal(self.get_label(target_entity))))
+                if mother_target:
+                    self.graph.add((rdflib.URIRef(source_entity), rdflib.RDFS.subClassOf, rdflib.URIRef(mother_target)))
 
             # Relation
             if self.check_entity(relation):

libabstractor/SparqlQuery.py

@@ -174,7 +174,7 @@ def process_query(self, query):
         list
             Parsed results
         """
-        # prefixed_query = self.get_sparl_prefix() + query
+        # query = self.get_sparl_prefix() + query
         logging.debug(query)
         if self.source_type == "sparql":
             return self.parse_sparql_results(self.execute_sparql_query(query))

setup.py

@@ -2,12 +2,12 @@
 
 setup(
     name='abstractor',
-    version='4.0.0',
+    version='4.1.0',
     description='Abstraction generator for AskOmics, from a distant SPARQL endpoint',
     author='Xavier Garnier',
     author_email='[email protected]',
     url='https://github.com/askomics/abstractor',
-    download_url='https://github.com/askomics/abstractor/archive/4.0.0.tar.gz',
+    download_url='https://github.com/askomics/abstractor/archive/4.1.0.tar.gz',
     install_requires=['SPARQLWrapper'],
     packages=find_packages(),
     license='AGPL',