Merge pull request #82 from monarch-initiative/58-docker-biolink-update

58 docker biolink update

Dockerfile

@@ -27,14 +27,20 @@ ENV JAVA_HOME /usr/lib/jvm/java-8-oracle
 # Define working directory.
 WORKDIR /data
 ADD owlsim-services/target/owlsim-services-3.0-SNAPSHOT.jar /data/
-ADD configuration-samples/configuration-all.yaml /data/configuration.yaml
+#ADD configuration-samples/configuration-all.yaml /data/configuration.yaml
+ADD scripts/golr-exporter.py /data/golr-exporter.py
+ADD scripts/configuration-generator.py /data/configuration-generator.py
 
-RUN if [ $species = "human" ];  \
+RUN apt-get -y update && apt-get install python-pip -y
+RUN pip install pyyaml
+
+RUN if [ $species = "all" ];  \
   then \
-    wget https://data.monarchinitiative.org/owl/all-hp.owl -O /data/all.owl; \
+    cd /data && python golr-exporter.py; \
   else \
-    wget http://ci.monarchinitiative.org/view/dev/job/create-owlsim-files-on-dev/lastSuccessfulBuild/artifact/server/all.owl -O /data/all.owl; \
+    cd /data && python golr-exporter.py -t $species; \
   fi
+RUN cd /data && python configuration-generator.py
 
 CMD java -jar /data/owlsim-services-3.0-SNAPSHOT.jar server /data/configuration.yaml
 

README.md

@@ -76,6 +76,7 @@ To restrict to specific species:
 
 ```
 docker build --build-arg species=all -t owlsim-all . # default if no args is provided or not matching any species
-docker build --build-arg species=human -t owlsim-human
-
+docker build --build-arg species=Hs -t owlsim-human .
+docker build --build-arg species=Hs,Mm -t owlsim-multi .
+# supported species: Hs,Mm,Dr,Dm,Ce
 ```

owlsim-core/src/main/java/org/monarchinitiative/owlsim/kb/impl/BMKnowledgeBaseOWLAPIImpl.java

@@ -363,8 +363,8 @@ private void setPropertyValues(OWLOntology ont, OWLNamedIndividual i) {
 						addPropertyValue(pvm, pid, getShortForm(((OWLNamedIndividual) obj).getIRI()));
 
 					}
-
-				} else if (false) {
+					// TODO check this...
+				} else if (false) { 
 					String pid = getShortForm(((OWLDataProperty) p).getIRI());
 					OWLLiteral obj = ((OWLDataPropertyAssertionAxiom) paa).getObject();
 					if (obj instanceof OWLLiteral) {

owlsim-services/src/main/java/org/monarchinitiative/owlsim/services/modules/KnowledgeBaseModule.java

@@ -33,7 +33,7 @@ public KnowledgeBaseModule(Collection<String> ontologyUris, Collection<String> o
 		logger.info("Loading curies:");
 		curies.entrySet().forEach(logger::info);
 		logger.info("Loading labels:");
-		//labels.entrySet().forEach(logger::info);
+		labelTsvs.forEach(logger::info);
 
 		//The OwlKnowledgeBase.Loader uses the ELKReasonerFactory and Concurrency.CONCURRENT as defaults.
 		this.bmKnowledgeBase = OwlKnowledgeBase.loader()

scripts/configuration-generator.py

@@ -0,0 +1,41 @@
+#!/usr/bin/python
+
+# grab the curie mapping
+# scan the current directory for the tsvs
+# output an owlsim config file
+
+import yaml
+import urllib2
+import fnmatch
+import os
+
+def main():
+
+    # Getting tsv file path
+    tsvs = []
+    for root, _, filenames in os.walk('.'):
+        for filename in fnmatch.filter(filenames, '*.tsv'):
+            tsvs.append(os.path.abspath(os.path.join(root, filename))) # Check that: it looks insane to get the absolute path...
+
+    dataTsvs = [filename for filename in tsvs if not filename.endswith('label.tsv')]
+    labelTsvs = [filename for filename in tsvs if filename.endswith('label.tsv')]
+
+    # Note: the yaml dump is buggy with the empty string key, so I'm just appending the raw string and add indentation manually
+    curies = urllib2.urlopen("https://raw.githubusercontent.com/monarch-initiative/dipper/master/dipper/curie_map.yaml").read()
+
+    data = dict(
+        ontologyUris = ['http://purl.obolibrary.org/obo/upheno/monarch.owl'],
+        ontologyDataUris = [],
+        dataTsvs = dataTsvs,
+        labelTsvs = labelTsvs
+    )
+
+    print("Generating owlsim configuration")
+    with open('configuration.yaml', 'w') as outfile:
+        yaml.dump(data, outfile, default_flow_style=False)
+        outfile.write("curies:\n")
+        outfile.write('  '.join(curies.splitlines(True)))
+
+
+if __name__ == "__main__":
+    main()

scripts/golr-exporter.py

@@ -43,7 +43,8 @@ def transformAssociation(input, output):
 
 def main():
 
-    biolink = "http://localhost:5000/api" #mart/labels/gene/phenotype/NCBITaxon%3A7955
+    #biolink = "http://localhost:5000/api" #mart/labels/gene/phenotype/NCBITaxon%3A7955
+    biolink = "https://api.monarchinitiative.org/api"
 
     taxon_map = {
         'Hs': 9606,