/**
* This class extracts the set of most suitable mediating ontologies for a given set of concept
* labels from ontology 1 and ontology 2
*
* @param representative_labels_onto1
* @param representative_labels_onto2
*/
public MediatingOntologyExtractor(Set<String> representative_labels) {
bioportal = new RESTBioPortalAccess();
if (bioportal.isActive()) {
// TODO To speed up process
// This could be commented since we already ignore those ontologies giving errors or not
// contributing to the composed mappings
bad_ontologies.add("OPE"); // gives an error
// bad_ontologies.add("SNOMEDCT"); //not available
bad_ontologies.add("NIFSTD"); // error when downloading it
bad_ontologies.add("RADLEX"); // interesting concepts only in individuals
// bad_ontologies.add("FMA"); //the download is a zip file and gives an error
bad_ontologies.add("SOPHARM"); // error downloading
bad_ontologies.add(
"EP"); // Basically it only contaisn terms from the Imported FMA, but is is imported from
// outside bioPortal (error downloading)
bad_ontologies.add("GO-PLUS"); // Error accessing metadata
this.representative_labels = representative_labels;
extractMediatingOntologies();
// if (Parameters.print_output_always)
printMediatingOntologies();
} else {
System.out.println("BioPotal is not accessible. No mediating ontology has been extracted.");
}
}
public OWLOntology downloadBioPortalOntology(String ontoAcronym, int attempts) {
return bioportal.downLoadOntology(ontoAcronym, attempts);
}
private void extractSuitableOntologiesForLabel(String label) {
try {
acronyms.clear();
int size_synset;
// TODO We do not require exact correspondence?
// If no exact we may get many crap and more than one page... time may is affected
// dramatically
JsonNode jsonConcepts = bioportal.getConcepts4Label(label, true);
// If an ontology appears several times for a concept
for (JsonNode concept : jsonConcepts) {
// TODO Something has change din BioPortal and no synonyms are given for UBERON
// if (concept.has(bioportal.SYNONYM)) {
String acronym =
bioportal.getOntologyAcronym(
concept.get(bioportal.LINKS).get(bioportal.ONTOLOGY).asText());
// Consider only one positive hit per label
// For example EHDA has 39 positive hits for only one label!!
// Also ignore ontologies that produce an error. E.g. have a wrong OWL format
if (acronyms.contains(acronym) || bad_ontologies.contains(acronym)) continue;
acronyms.add(acronym);
size_synset = 0;
if (concept.has(bioportal.SYNONYM)) size_synset = concept.get(bioportal.SYNONYM).size();
// System.out.println("\t"+acronym);
// Already in list
if (candiateMOs.containsKey(acronym)) {
candiateMOs.get(acronym).increasePositiveHits();
candiateMOs.get(acronym).increaseNumberOfProvidedSynonyms(size_synset);
} else {
// We call bioportal about metrics
BioPortalOntologyInfo onto_bio = bioportal.getMetricInfo4Onto(acronym);
/// if (onto_bio.getClasses()<=MAX_SIZE_ONTOLOGY){
candiateMOs.put(acronym, new CandidateMediatingOntology(onto_bio, size_synset));
// }
}
// }
}
} catch (JsonProcessingException e) {
// TODO Auto-generated catch block
System.out.println(
"Error performing the BioPortal call for: " + label + ". Mediating ontology search.");
} catch (IOException e) {
// TODO Auto-generated catch block
System.out.println(
"Error performing the BioPortal call for: " + label + ". Mediating ontology search.");
}
}
