Merge pull request #71 from julesjacobsen/master

Adding ability to load individuals from in-memory data as well as off-disk.

owlsim-core/src/main/java/org/monarchinitiative/owlsim/io/Ontology.java

@@ -17,6 +17,7 @@
 import java.util.Map;
 import java.util.Objects;
 import java.util.Set;
+import java.util.function.BiConsumer;
 import java.util.zip.GZIPInputStream;
 
 /**
@@ -34,13 +35,15 @@
     private final OWLOntology owlOntology;
 
     private final OWLOntologyManager ontologyManager;
+    private final OWLDataFactory owlDataFactory;
 
     private Ontology(OntologySourceData sourceData, Concurrency concurrency) {
         Objects.requireNonNull(sourceData, "Unable to create Ontology without data sources.");
         this.sourceData = sourceData;
         this.curieUtil = new CurieUtil(sourceData.getCuries());
         this.ontologyManager = createOntologyManager(concurrency);
         this.owlOntology = createEmptyOntology(ontologyManager);
+        this.owlDataFactory = ontologyManager.getOWLDataFactory();
         loadOwlOntology();
     }
 
@@ -81,12 +84,29 @@ public CurieUtil getCurieUtil() {
         return curieUtil;
     }
 
+    /**
+     * @param curie
+     * @return
+     */
+    public OWLClass getOWLClass(String curie) {
+        return owlDataFactory.getOWLClass(toIri(curie));
+    }
+
+    public OWLNamedIndividual getOWLNamedIndividual(String curie) {
+        return owlDataFactory.getOWLNamedIndividual(toIri(curie));
+    }
+
+    public String toCurie(IRI iri) {
+        String iriString = iri.toString();
+        return curieUtil.getCurie(iriString).orElse(iriString);
+    }
+
     private void loadOwlOntology() {
         //Order matters here - don't change it.
         mergeOntologies(sourceData.getOntologies());
         mergeOntologies(sourceData.getDataOntologies());
-        loadDataFromTsv(sourceData.getDataTsvs());
-        loadDataFromPairwiseMappings(sourceData.getPairwiseMappings());
+        loadIndividualAssociationsFromTsv(sourceData.getIndividualAssociationTsvs());
+        loadIndividualAssociationsFromMap(sourceData.getIndividualAssociations());
         logger.info("Ontology loaded");
     }
 
@@ -118,10 +138,6 @@ private OWLOntology mergeOntologies(Collection<String> uris) {
         return owlOntology;
     }
 
-    private ChangeApplied addAxiom(OWLAxiom axiom) {
-        return ontologyManager.addAxiom(owlOntology, axiom);
-    }
-
     private ChangeApplied addAxioms(Set<OWLAxiom> axioms) {
         return ontologyManager.addAxioms(owlOntology, axioms);
     }
@@ -159,12 +175,12 @@ private OWLOntology loadOwlOntology(IRI iri) {
         }
     }
 
-    private OWLOntology loadDataFromTsv(Collection<String> paths) {
-        paths.forEach(this::loadDataFromTsv);
+    private OWLOntology loadIndividualAssociationsFromTsv(Collection<String> paths) {
+        paths.forEach(this::loadIndividualAssociationsFromTsv);
         return owlOntology;
     }
 
-    private OWLOntology loadDataFromTsv(String path) {
+    private OWLOntology loadIndividualAssociationsFromTsv(String path) {
         if (path.endsWith(".gz")) {
             return loadDataFromTsvGzip(path);
         }
@@ -191,8 +207,20 @@ private OWLOntology loadDataFromTsvGzip(String path) {
         return owlOntology;
     }
 
-    private void loadDataFromPairwiseMappings(Map<String, String> pairwiseMappings) {
-        pairwiseMappings.forEach(this::addInstanceOf);
+    private void loadIndividualAssociationsFromMap(Map<String, Collection<String>> individuals) {
+        if(!individuals.isEmpty()){
+            logger.info("Loading individuals from map");
+        }
+        //e.g. 'ORPHA:710': ['HP:0000194','HP:0000218','HP:0000262','HP:0000303','HP:0000316']
+        individuals.forEach(addIndividual());
+    }
+
+    private BiConsumer<String, Collection<String>> addIndividual() {
+        return (individual, annotations) -> {
+            for (String curie : annotations) {
+                addInstanceOf(individual, curie);
+            }
+        };
     }
 
     private void loadLineIntoDataOntology(String line) {
@@ -204,15 +232,22 @@ private void loadLineIntoDataOntology(String line) {
     }
 
     private void addInstanceOf(String individual, String ontologyClass) {
-//                logger.info("Adding axiom " + individual + ": " + ontologyClass);
-        OWLDataFactory owlDataFactory = ontologyManager.getOWLDataFactory();
-        OWLClass owlClass = owlDataFactory.getOWLClass(toIri(ontologyClass));
-        OWLNamedIndividual owlNamedIndividual = owlDataFactory.getOWLNamedIndividual(toIri(individual));
-        OWLClassAssertionAxiom axiom = owlDataFactory.getOWLClassAssertionAxiom(owlClass, owlNamedIndividual);
-        addAxiom(axiom);
+        Objects.requireNonNull(individual, "Individual identifier cannot be null. Check your input.");
+        Objects.requireNonNull(ontologyClass, "Class identifier(s) cannot be null. Check your input.");
+        if (!ontologyClass.isEmpty()) {
+//            logger.info("Adding axiom " + individual + ": " + ontologyClass);
+            OWLClass owlClass = getOWLClass(ontologyClass);
+            OWLNamedIndividual owlNamedIndividual = getOWLNamedIndividual(individual);
+            OWLClassAssertionAxiom axiom = owlDataFactory.getOWLClassAssertionAxiom(owlClass, owlNamedIndividual);
+            addAxiom(axiom);
+        }
+    }
+
+    private ChangeApplied addAxiom(OWLAxiom axiom) {
+        return ontologyManager.addAxiom(owlOntology, axiom);
     }
 
-    private IRI toIri(String id) {
+    IRI toIri(String id) {
         return IRI.create(curieUtil.getIri(id).orElse(id));
     }
 

owlsim-core/src/main/java/org/monarchinitiative/owlsim/io/OntologySourceData.java

@@ -20,16 +20,16 @@
     private final List<String> dataOntologies;
 
     private final Map<String, String> curies;
-    private final List<String> dataTsvs;
-    //TODO: add these so people can programmatically add individual assertions
-    private final Map<String, String> pairwiseMappings;
+    private final List<String> individualAssociationTsvs;
+    private final Map<String, Collection<String>> individualAssociations;
+    //TODO - labels?
 
     private OntologySourceData(Builder builder) {
         this.ontologies = distinctImmutableListOf(builder.ontologies);
         this.dataOntologies = distinctImmutableListOf(builder.dataOntologies);
         this.curies = ImmutableMap.copyOf(builder.curies);
-        this.dataTsvs = distinctImmutableListOf(builder.dataTsvs);
-        this.pairwiseMappings = ImmutableMap.copyOf(builder.pairwiseMappings);
+        this.individualAssociationTsvs = distinctImmutableListOf(builder.individualAssociationTsvs);
+        this.individualAssociations = ImmutableMap.copyOf(builder.individualAssociations);
     }
 
     private ImmutableList<String> distinctImmutableListOf(List<String> list) {
@@ -48,12 +48,12 @@ private OntologySourceData(Builder builder) {
         return curies;
     }
 
-    public List<String> getDataTsvs() {
-        return dataTsvs;
+    public List<String> getIndividualAssociationTsvs() {
+        return individualAssociationTsvs;
     }
 
-    public Map<String, String> getPairwiseMappings() {
-        return pairwiseMappings;
+    public Map<String, Collection<String>> getIndividualAssociations() {
+        return individualAssociations;
     }
 
     @Override
@@ -64,13 +64,13 @@ public boolean equals(Object o) {
         return Objects.equals(ontologies, that.ontologies) &&
                 Objects.equals(dataOntologies, that.dataOntologies) &&
                 Objects.equals(curies, that.curies) &&
-                Objects.equals(dataTsvs, that.dataTsvs) &&
-                Objects.equals(pairwiseMappings, that.pairwiseMappings);
+                Objects.equals(individualAssociationTsvs, that.individualAssociationTsvs) &&
+                Objects.equals(individualAssociations, that.individualAssociations);
     }
 
     @Override
     public int hashCode() {
-        return Objects.hash(ontologies, dataOntologies, curies, dataTsvs, pairwiseMappings);
+        return Objects.hash(ontologies, dataOntologies, curies, individualAssociationTsvs, individualAssociations);
     }
 
     @Override
@@ -79,8 +79,8 @@ public String toString() {
                 "ontologies=" + ontologies +
                 ", dataOntologies=" + dataOntologies +
                 ", curies=" + curies +
-                ", dataTsvs=" + dataTsvs +
-                ", pairwiseMappings=" + pairwiseMappings +
+                ", individualAssociationTsvs=" + individualAssociationTsvs +
+                ", individualAssociations=" + individualAssociations +
                 '}';
     }
 
@@ -93,8 +93,8 @@ public static Builder builder() {
         private List<String> dataOntologies = new ArrayList<>();
         //Curies need to be supplied if people are adding data using TSV files or pairwise mappings using curies.
         private Map<String, String> curies = Collections.emptyMap();
-        private List<String> dataTsvs = new ArrayList<>();
-        private Map<String, String> pairwiseMappings = Collections.emptyMap();
+        private List<String> individualAssociationTsvs = new ArrayList<>();
+        private Map<String, Collection<String>> individualAssociations = new HashMap<>();
 
         private Builder(){
             //use the static method.
@@ -161,33 +161,38 @@ public Builder dataOntologies(Collection<String> paths) {
             return this;
         }
 
-        public Builder dataTsv(String path) {
-            dataTsvs.add(path);
+        public Builder individualAssociationsTsv(String path) {
+            individualAssociationTsvs.add(path);
             return this;
         }
 
-        public Builder dataTsv(String... paths) {
-            dataTsvs.addAll(Arrays.asList(paths));
+        public Builder individualAssociationsTsv(String... paths) {
+            individualAssociationTsvs.addAll(Arrays.asList(paths));
             return this;
         }
 
-        public Builder dataTsv(Collection<String> paths) {
-            dataTsvs.addAll(paths);
+        public Builder individualAssociationsTsv(Collection<String> paths) {
+            individualAssociationTsvs.addAll(paths);
+            return this;
+        }
+
+        public Builder individualAssociations(Map<String, ? extends Collection<String>> mappings) {
+            individualAssociations.putAll(mappings);
             return this;
         }
 
         public OntologySourceData build() {
             if(ontologies.isEmpty()) {
                 throw new OntologyLoadException("No ontology defined.");
             }
-            if (curies.isEmpty() && hasNonOntologyData()) {
-                throw new OntologyLoadException("Cannot load TSV data sources or pairwise mappings when curies have not been defined.");
+            if (curies.isEmpty() && hasIndividualAssociationData()) {
+                throw new OntologyLoadException("Cannot load individual class associations when curies have not been defined.");
             }
             return new OntologySourceData(this);
         }
 
-        private boolean hasNonOntologyData() {
-            return !dataTsvs.isEmpty() || !pairwiseMappings.isEmpty();
+        private boolean hasIndividualAssociationData() {
+            return !individualAssociationTsvs.isEmpty() || !individualAssociations.isEmpty();
         }
     }
 }

owlsim-core/src/main/java/org/monarchinitiative/owlsim/io/OwlKnowledgeBase.java

@@ -57,7 +57,6 @@ public Loader loadCuries(Map<String, String> curies) {
          * @param file
          */
         public Loader loadOntology(File file) {
-            Math.random();
             sourceDataBuilder.ontology(file);
             return this;
         }
@@ -103,18 +102,23 @@ public Loader loadDataFromOntologies(Collection<String> paths) {
             return this;
         }
 
-        public Loader loadDataFromTsv(String path) {
-            sourceDataBuilder.dataTsv(path);
+        public Loader loadIndividualAssociationsFromTsv(String path) {
+            sourceDataBuilder.individualAssociationsTsv(path);
             return this;
         }
 
-        public Loader loadDataFromTsv(String... paths) {
-            sourceDataBuilder.dataTsv(paths);
+        public Loader loadIndividualAssociationsFromTsv(String... paths) {
+            sourceDataBuilder.individualAssociationsTsv(paths);
             return this;
         }
 
-        public Loader loadDataFromTsv(Collection<String> paths) {
-            sourceDataBuilder.dataTsv(paths);
+        public Loader loadIndividualAssociationsFromTsv(Collection<String> paths) {
+            sourceDataBuilder.individualAssociationsTsv(paths);
+            return this;
+        }
+
+        public Loader loadIndividualAssociations(Map<String, ? extends Collection<String>> data) {
+            sourceDataBuilder.individualAssociations(data);
             return this;
         }
 

owlsim-core/src/main/java/org/monarchinitiative/owlsim/kb/BMKnowledgeBase.java

@@ -1,13 +1,13 @@
 package org.monarchinitiative.owlsim.kb;
 
-import java.util.Map;
-import java.util.Set;
-
+import com.googlecode.javaewah.EWAHCompressedBitmap;
+import org.monarchinitiative.owlsim.io.OwlKnowledgeBase;
 import org.monarchinitiative.owlsim.kb.impl.BMKnowledgeBaseOWLAPIImpl;
 import org.monarchinitiative.owlsim.model.kb.Attribute;
 import org.monarchinitiative.owlsim.model.kb.Entity;
 
-import com.googlecode.javaewah.EWAHCompressedBitmap;
+import java.util.Map;
+import java.util.Set;
 
 /**
  * An interface to an ontology in which the fundamental unit of representation of 
@@ -351,7 +351,9 @@
 	public int getRootIndex();
 
 
-
+	public static OwlKnowledgeBase.Loader owlLoader() {
+		return OwlKnowledgeBase.loader();
+	}