/**
* 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;
}
@Test
public void testMarshalPhobiusMatch() {
init();
final Set<PhobiusMatch.PhobiusLocation> locations = new HashSet<PhobiusMatch.PhobiusLocation>();
locations.add(new PhobiusMatch.PhobiusLocation(1, 2));
final Signature signature =
new Signature.Builder("SIGNAL_PEPTIDE").name("Signal Peptide").build();
final PhobiusMatch match = new PhobiusMatch(signature, locations);
final Protein protein = new Protein("aaa");
protein.addMatch(match);
try {
final String result = marshal(marshaller, protein);
if (LOGGER.isDebugEnabled()) {
LOGGER.debug(result);
}
assertNotNull("XML result shouldn't be null!", result);
assertTrue(result.contains("<matches>"));
assertTrue(result.contains("<phobius-match>"));
assertTrue(result.contains("<signature"));
assertTrue(result.contains("name=\"Signal Peptide\""));
assertTrue(result.contains("ac=\"SIGNAL_PEPTIDE\""));
assertTrue(result.contains("<locations>"));
assertTrue(result.contains("<phobius-location"));
assertTrue(result.contains("end=\"2\""));
assertTrue(result.contains("start=\"1\""));
} catch (IOException e) {
LOGGER.warn("Couldn't marshal protein object!", e);
}
}
/**
* 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;
}
/**
* Writes out protein view to an zipped and compressed HTML file.
*
* @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(final Protein protein) throws IOException {
checkEntryHierarchy();
if (entryHierarchy != null) {
for (ProteinXref xref : protein.getCrossReferences()) {
final SimpleProtein simpleProtein = SimpleProtein.valueOf(protein, xref, entryHierarchy);
if (simpleProtein != null) {
// Build model for FreeMarker
final SimpleHash model = buildModelMap(simpleProtein, entryHierarchy);
// Render template and write result to a file
Writer writer = null;
try {
final Template temp = freeMarkerConfig.getTemplate(freeMarkerTemplate);
checkTempDirectory(tempDirectory);
if (!tempDirectory.endsWith("/")) {
tempDirectory = tempDirectory + "/";
}
UrlFriendlyIdGenerator gen = UrlFriendlyIdGenerator.getInstance();
String urlFriendlyId = gen.generate(xref.getIdentifier());
final File newResultFile = new File(tempDirectory + urlFriendlyId + ".svg");
resultFiles.add(newResultFile);
writer = new PrintWriter(new FileWriter(newResultFile));
temp.process(model, writer);
writer.flush();
} catch (TemplateException e) {
e.printStackTrace();
} catch (IOException e) {
e.printStackTrace();
} finally {
if (writer != null) {
writer.close();
}
}
}
}
}
return 0;
}
/**
* Inserts new Proteins. If there are Protein objects with the same MD5 / sequence in the
* database, this method updates these proteins, rather than inserting the new ones.
*
* <p>Note that this method inserts the new Protein objects AND and new Xrefs (possibly updating
* an existing Protein object if necessary with the new Xref.)
*
* @param newProteins being a List of new Protein objects to insert
* @return a new List<Protein> containing all of the inserted / updated Protein objects. (Allows
* the caller to retrieve the primary keys for the proteins).
*/
@Transactional
@SuppressWarnings("unchecked")
public PersistedProteins insertNewProteins(Collection<Protein> newProteins) {
PersistedProteins persistentProteins = new PersistedProteins();
if (newProteins.size() > 0) {
// Create a List of MD5s (just as Strings) to query the database with
final List<String> newMd5s = new ArrayList<String>(newProteins.size());
for (Protein newProtein : newProteins) {
newMd5s.add(newProtein.getMd5());
if (LOGGER.isDebugEnabled()) {
LOGGER.debug("MD5 of new protein: " + newProtein.getMd5());
}
}
// Retrieve any proteins AND associated xrefs that have the same MD5 as one of the 'new'
// proteins
// being inserted and place in a Map of MD5 to Protein object.
final Map<String, Protein> md5ToExistingProtein = new HashMap<String, Protein>();
final Query query =
entityManager.createQuery(
"select p from Protein p left outer join fetch P.crossReferences where p.md5 in (:md5)");
query.setParameter("md5", newMd5s);
for (Protein existingProtein : (List<Protein>) query.getResultList()) {
if (LOGGER.isDebugEnabled()) {
LOGGER.debug("Found 1 existing protein with MD5: " + existingProtein.getMd5());
}
md5ToExistingProtein.put(existingProtein.getMd5(), existingProtein);
}
// Now have the List of 'new' proteins, and a list of existing proteins that match
// them. Insert / update proteins as appropriate.
for (Protein candidate : newProteins) {
// PROTEIN ALREADY EXISTS in the DB. - update cross references and save.
if (md5ToExistingProtein.keySet().contains(candidate.getMd5())) {
// This protein is already in the database - add any new Xrefs and update.
Protein existingProtein = md5ToExistingProtein.get(candidate.getMd5());
boolean updateRequired = false;
if (candidate.getCrossReferences() != null) {
if (LOGGER.isTraceEnabled()) {
LOGGER.trace("Protein TO BE STORED has xrefs:");
}
for (ProteinXref xref : candidate.getCrossReferences()) {
if (LOGGER.isTraceEnabled()) {
LOGGER.trace(xref.getIdentifier());
}
// Add any NEW cross references.
if (!existingProtein.getCrossReferences().contains(xref)) {
if (LOGGER.isTraceEnabled()) {
LOGGER.trace(
"Adding " + xref.getIdentifier() + " and setting updateRequired = true");
}
existingProtein.addCrossReference(xref);
updateRequired = true;
}
}
}
if (updateRequired) {
// PROTEIN is NOT new, but CHANGED (new Xrefs)
if (LOGGER.isTraceEnabled()) {
LOGGER.trace("Merging protein with new Xrefs: " + existingProtein.getMd5());
}
entityManager.merge(existingProtein);
}
persistentProteins.addPreExistingProtein(existingProtein);
}
// PROTEIN IS NEW - save it.
else {
if (LOGGER.isTraceEnabled()) {
LOGGER.trace("Saving new protein: " + candidate.getMd5());
}
entityManager.persist(candidate);
persistentProteins.addNewProtein(candidate);
// Check for this new protein next time through the loop, just in case the new source of
// proteins is redundant (e.g. a FASTA file with sequences repeated).
md5ToExistingProtein.put(candidate.getMd5(), candidate);
}
}
}
// Finally return all the persisted Protein objects (new or existing)
entityManager.flush();
return persistentProteins;
}