/**
* Retrieves all Proteins, cross references and matches for a range
*
* @param bottom range lower bound (included)
* @param top range upper bound (included)
* @return The Protein, with matches loaded. (matches are LAZY by default) or null if the primary
* key is not present in the database.
*/
@Override
@Transactional(readOnly = true)
@SuppressWarnings("unchecked")
public List<Protein> getProteinsAndMatchesAndCrossReferencesBetweenIds(long bottom, long top) {
Query query =
entityManager.createQuery(
"select distinct p from Protein p "
+ "left outer join fetch p.matches "
+ "left outer join fetch p.crossReferences where p.id >= :bottom and p.id <= :top");
query.setParameter("bottom", bottom);
query.setParameter("top", top);
List<Protein> matchingProteins = (List<Protein>) query.getResultList();
if (LOGGER.isTraceEnabled()) {
LOGGER.trace("Querying proteins with IDs in range: " + bottom + " to " + top);
LOGGER.trace("Matching protein count: " + matchingProteins.size());
for (Protein protein : matchingProteins) {
LOGGER.trace("Protein ID: " + protein.getId() + " MD5: " + protein.getMd5());
LOGGER.trace("Has " + protein.getMatches().size() + " matches");
for (ProteinXref xref : protein.getCrossReferences()) {
LOGGER.trace("Xref: " + xref.getIdentifier());
}
}
}
return matchingProteins;
}
/**
* Writes out all protein matches for the specified protein (GFF formatted).
*
* @param protein containing matches to be written out
* @return the number of rows printed (i.e. the number of Locations on Matches).
* @throws java.io.IOException in the event of I/O problem writing out the file.
*/
public int write(Protein protein) throws IOException {
List<String> proteinIdsForGFF = getProteinAccessions(protein);
int sequenceLength = protein.getSequenceLength();
String md5 = protein.getMd5();
String date = dmyFormat.format(new Date());
Set<Match> matches = protein.getMatches();
// Write sequence region information
for (String proteinIdForGFF : proteinIdsForGFF) {
if (matches.size() > 0) {
// Check if protein accessions are GFF3 valid
proteinIdForGFF = ProteinMatchesGFFResultWriter.getValidGFF3SeqId(proteinIdForGFF);
// Write sequence-region
super.gffWriter.write("##sequence-region " + proteinIdForGFF + " 1 " + sequenceLength);
if (writeFullGFF) {
writeReferenceLine(proteinIdForGFF, sequenceLength, md5);
addFASTASeqToMap(proteinIdForGFF, protein.getSequence());
}
processMatches(matches, proteinIdForGFF, date, protein, proteinIdForGFF, writeFullGFF);
} // end match size check
}
return 0;
}