/**
* Resolve Gene primaryId from a symbol.
*
* @param primaryId the fly gene primaryId
* @param symbol the fly gene symbol
* @return gene primaryId
* @throws ObjectStoreException
*/
private String createGene(String primaryId, String symbol) throws ObjectStoreException {
if (primaryId == null) { // for miRNA case
if ("mlc-c_in1".endsWith(symbol)) {
symbol = "Mlc-c";
}
if ("mir-iab-4as".endsWith(symbol)) {
symbol = "mir-iab-8";
}
IdResolver resolver = resolverFactory.getIdResolver();
int resCount = resolver.countResolutions(FLY_TAXON_ID, symbol);
if (resCount != 1) {
LOG.info(
"RESOLVER: failed to resolve gene to one identifier, ignoring gene: "
+ symbol
+ " count: "
+ resCount
+ " FBgn: "
+ resolver.resolveId(FLY_TAXON_ID, symbol));
}
primaryId = resolver.resolveId(FLY_TAXON_ID, symbol).iterator().next();
}
if (!geneItems.containsKey(primaryId)) {
createBioEntity("Gene", primaryId, symbol);
}
return primaryId;
}
private String getGene(String identifierType, String id, String taxonId)
throws ObjectStoreException {
String identifier = id;
if (rslv != null && rslv.hasTaxon(taxonId)) {
identifier = resolveGene(identifier, taxonId);
if (identifier == null) {
return null;
}
}
String refId = identifiersToGenes.get(identifier);
if (refId == null) {
Item gene = createItem("Gene");
refId = gene.getIdentifier();
gene.setAttribute(identifierType, identifier);
gene.setReference("organism", getOrganism(taxonId));
identifiersToGenes.put(identifier, refId);
try {
store(gene);
} catch (ObjectStoreException e) {
throw new ObjectStoreException(e);
}
}
return refId;
}
private String resolveGene(String originalId, String taxonId) {
String primaryIdentifier = null;
int resCount = rslv.countResolutions(taxonId, originalId);
if (resCount != 1) {
LOG.info(
"RESOLVER: failed to resolve gene to one identifier, ignoring "
+ "gene: "
+ originalId
+ " for organism "
+ taxonId
+ " count: "
+ resCount
+ " found ids: "
+ rslv.resolveId(taxonId, originalId)
+ ".");
} else {
primaryIdentifier = rslv.resolveId(taxonId, originalId).iterator().next();
LOG.info("RESOLVER found gene " + primaryIdentifier + " for original id: " + originalId);
}
return primaryIdentifier;
}
private String resolveTerm(String identifier) {
String goId = identifier;
IdResolver resolver = ontologyResolverFactory.getIdResolver(false);
if (resolver != null) {
int resCount = resolver.countResolutions("0", identifier);
if (resCount > 1) {
LOG.info(
"RESOLVER: failed to resolve ontology term to one identifier, "
+ "ignoring term: "
+ identifier
+ " count: "
+ resCount
+ " : "
+ resolver.resolveId("0", identifier));
return null;
}
if (resCount == 1) {
goId = resolver.resolveId("0", identifier).iterator().next();
}
}
return goId;
}
private String newProduct(
String identifier,
String type,
Item organism,
String dataSourceCode,
boolean createOrganism,
String field)
throws ObjectStoreException {
String idField = field;
String accession = identifier;
String clsName = null;
// find gene attribute first to see if organism should be part of key
if ("gene".equalsIgnoreCase(type)) {
clsName = "Gene";
String taxonId = organism.getAttribute("taxonId").getValue();
if (idField == null) {
idField = configs.get(taxonId).identifier;
if (idField == null) {
throw new RuntimeException(
"Could not find a identifier property for taxon: "
+ taxonId
+ " check properties file: "
+ PROP_FILE);
}
}
// if a Dmel gene we need to use FlyBaseIdResolver to find a current id
if ("7227".equals(taxonId)) {
IdResolver resolver = flybaseResolverFactory.getIdResolver(false);
if (resolver != null) {
int resCount = resolver.countResolutions(taxonId, accession);
if (resCount != 1) {
LOG.info(
"RESOLVER: failed to resolve gene to one identifier, "
+ "ignoring gene: "
+ accession
+ " count: "
+ resCount
+ " FBgn: "
+ resolver.resolveId(taxonId, accession));
return null;
}
accession = resolver.resolveId(taxonId, accession).iterator().next();
}
}
} else if ("protein".equalsIgnoreCase(type)) {
// TODO use values in config
clsName = "Protein";
idField = "primaryAccession";
} else {
String typeCls = TypeUtil.javaiseClassName(type);
if (getModel().getClassDescriptorByName(typeCls) != null) {
Class<?> cls = getModel().getClassDescriptorByName(typeCls).getType();
if (BioEntity.class.isAssignableFrom(cls)) {
clsName = typeCls;
}
}
if (clsName == null) {
throw new IllegalArgumentException("Unrecognised annotation type '" + type + "'");
}
}
boolean includeOrganism;
if ("primaryIdentifier".equals(idField) || "protein".equals(type)) {
includeOrganism = false;
} else {
includeOrganism = createOrganism;
}
String key = makeProductKey(accession, type, organism, includeOrganism);
// Have we already seen this product somewhere before?
// if so, return the product rather than creating a new one...
if (productMap.containsKey(key)) {
return productMap.get(key);
}
// if a Dmel gene we need to use FlyBaseIdResolver to find a current id
Item product = createItem(clsName);
if (organism != null && createOrganism) {
product.setReference("organism", organism.getIdentifier());
}
product.setAttribute(idField, accession);
String dataSetIdentifier = getDataset(dataSourceCode);
product.addToCollection("dataSets", dataSetIdentifier);
Integer storedProductId = store(product);
storedProductIds.put(product.getIdentifier(), storedProductId);
productMap.put(key, product.getIdentifier());
return product.getIdentifier();
}
