downloader: add MONDO ontology

opencb · Jan 2, 2024 · 0004be6 · 0004be6
1 parent ce1767a
commit 0004be6
Show file tree

Hide file tree

Showing 4 changed files with 22 additions and 0 deletions.
diff --git a/cellbase-core/src/main/java/org/opencb/cellbase/core/config/DownloadProperties.java b/cellbase-core/src/main/java/org/opencb/cellbase/core/config/DownloadProperties.java
@@ -71,6 +71,7 @@ public class DownloadProperties {
     private URLProperties hpoObo;
     private URLProperties goObo;
     private URLProperties doidObo;
+    private URLProperties mondoObo;
     private URLProperties goAnnotation;
     private URLProperties revel;
     private URLProperties pubmed;
@@ -527,6 +528,15 @@ public DownloadProperties setCancerHotspot(URLProperties cancerHotspot) {
         return this;
     }
 
+    public URLProperties getMondoObo() {
+        return mondoObo;
+    }
+
+    public DownloadProperties setMondoObo(URLProperties mondoObo) {
+        this.mondoObo = mondoObo;
+        return this;
+    }
+
     public static class EnsemblProperties {
 
         private DatabaseCredentials database;

diff --git a/cellbase-core/src/main/resources/configuration.yml b/cellbase-core/src/main/resources/configuration.yml
@@ -177,6 +177,8 @@ download:
     host: http://purl.obolibrary.org/obo/go/go-basic.obo
   doidObo:
     host: http://purl.obolibrary.org/obo/doid.obo
+  mondoObo:
+    host: http://purl.obolibrary.org/obo/mondo.obo
   goAnnotation:
     host: http://geneontology.org/gene-associations/goa_human.gaf.gz
   revel:

diff --git a/cellbase-lib/src/main/java/org/opencb/cellbase/lib/EtlCommons.java b/cellbase-lib/src/main/java/org/opencb/cellbase/lib/EtlCommons.java
@@ -71,6 +71,7 @@ public class EtlCommons {
     public static final String HPO_FILE = "hp.obo";
     public static final String GO_FILE = "go-basic.obo";
     public static final String DOID_FILE = "doid.obo";
+    public static final String MONDO_FILE = "mondo.obo";
     public static final String PFM_DATA = "regulatory_pfm";
 
     // Build specific data options

diff --git a/cellbase-lib/src/main/java/org/opencb/cellbase/lib/builders/OntologyBuilder.java b/cellbase-lib/src/main/java/org/opencb/cellbase/lib/builders/OntologyBuilder.java
@@ -32,12 +32,14 @@ public class OntologyBuilder extends CellBaseBuilder {
     private Path hpoFile;
     private Path goFile;
     private Path doidFile;
+    private Path mondoFile;
 
     public OntologyBuilder(Path oboDirectoryPath, CellBaseSerializer serializer) {
         super(serializer);
         hpoFile = oboDirectoryPath.resolve(EtlCommons.HPO_FILE);
         goFile = oboDirectoryPath.resolve(EtlCommons.GO_FILE);
         doidFile = oboDirectoryPath.resolve(EtlCommons.DOID_FILE);
+        mondoFile = oboDirectoryPath.resolve(EtlCommons.MONDO_FILE);
     }
 
     @Override
@@ -64,6 +66,13 @@ public void parse() throws Exception {
             serializer.serialize(term);
         }
 
+        bufferedReader = FileUtils.newBufferedReader(mondoFile);
+        terms = parser.parseOBO(bufferedReader, "Mondo Ontology");
+        for (OntologyTerm term : terms) {
+            term.setSource("MONDO");
+            serializer.serialize(term);
+        }
+
         serializer.close();
     }
 }