private void readFile(String blastFile) throws IOException, ParseException {
logger.info("Start reading " + blastFile);
try {
blastDoc = XMLHelper.loadXML(blastFile);
} catch (SAXException ex) {
logger.error("A parsing error has occurred while reading XML blast file");
throw new ParseException(ex.getMessage(), 0);
} catch (ParserConfigurationException ex) {
logger.error("Internal XML parser non properly configured");
throw new ParseException(ex.getMessage(), 0);
}
logger.info("Read finished");
}
public LinkedHashMap<String, DNASequence> getDNACodingSequences() throws Exception {
LinkedHashMap<String, DNASequence> dnaSequenceList = new LinkedHashMap<String, DNASequence>();
ArrayList<Element> elementList =
XMLHelper.selectElements(geneidDoc.getDocumentElement(), "prediction/gene/cDNA");
logger.info("{} hits", elementList.size());
for (Element dnaElement : elementList) {
Element geneElement = (Element) dnaElement.getParentNode();
String sequence = dnaElement.getTextContent().replaceAll("\\W", "");
DNASequence dnaSequence = new DNASequence(sequence);
String idGene = geneElement.getAttribute("idGene");
dnaSequence.setAccession(new AccessionID(idGene));
dnaSequenceList.put(idGene, dnaSequence);
}
return dnaSequenceList;
}
public LinkedHashMap<String, ProteinSequence> getProteinSequences() throws Exception {
LinkedHashMap<String, ProteinSequence> proteinSequenceList =
new LinkedHashMap<String, ProteinSequence>();
ArrayList<Element> elementList =
XMLHelper.selectElements(geneidDoc.getDocumentElement(), "prediction/gene/protein");
logger.info("{} hits", elementList.size());
for (Element proteinElement : elementList) {
Element geneElement = (Element) proteinElement.getParentNode();
String sequence = proteinElement.getTextContent().replaceAll("\\W", "");
ProteinSequence proteinSequence = new ProteinSequence(sequence);
String idGene = geneElement.getAttribute("idGene");
proteinSequence.setAccession(new AccessionID(idGene));
proteinSequenceList.put(idGene, proteinSequence);
}
return proteinSequenceList;
}
public List<Result> createObjects(double maxEScore) throws IOException, ParseException {
if (targetFile == null) throw new IllegalStateException("File to be parsed not specified.");
// getAbsolutePath throws SecurityException
readFile(targetFile.getAbsolutePath());
// create mappings between sequences and blast id
mapIds();
ArrayList<Result> resultsCollection;
ArrayList<Hit> hitsCollection;
ArrayList<Hsp> hspsCollection;
try {
// select top level elements
String program =
XMLHelper.selectSingleElement(blastDoc.getDocumentElement(), "BlastOutput_program")
.getTextContent();
String version =
XMLHelper.selectSingleElement(blastDoc.getDocumentElement(), "BlastOutput_version")
.getTextContent();
String reference =
XMLHelper.selectSingleElement(blastDoc.getDocumentElement(), "BlastOutput_reference")
.getTextContent();
String dbFile =
XMLHelper.selectSingleElement(blastDoc.getDocumentElement(), "BlastOutput_db")
.getTextContent();
logger.info("Query for hits in " + targetFile);
ArrayList<Element> IterationsList =
XMLHelper.selectElements(
blastDoc.getDocumentElement(), "BlastOutput_iterations/Iteration[Iteration_hits]");
logger.info(IterationsList.size() + " results");
resultsCollection = new ArrayList<Result>();
for (Element element : IterationsList) {
BlastResultBuilder resultBuilder = new BlastResultBuilder();
// will add BlastOutput* key sections in the result object
resultBuilder
.setProgram(program)
.setVersion(version)
.setReference(reference)
.setDbFile(dbFile);
// Iteration* section keys:
resultBuilder
.setIterationNumber(
new Integer(
XMLHelper.selectSingleElement(element, "Iteration_iter-num").getTextContent()))
.setQueryID(
XMLHelper.selectSingleElement(element, "Iteration_query-ID").getTextContent())
.setQueryDef(
XMLHelper.selectSingleElement(element, "Iteration_query-def").getTextContent())
.setQueryLength(
new Integer(
XMLHelper.selectSingleElement(element, "Iteration_query-len")
.getTextContent()));
if (queryReferences != null)
resultBuilder.setQuerySequence(
queryReferencesMap.get(
XMLHelper.selectSingleElement(element, "Iteration_query-ID").getTextContent()));
Element iterationHitsElement = XMLHelper.selectSingleElement(element, "Iteration_hits");
ArrayList<Element> hitList = XMLHelper.selectElements(iterationHitsElement, "Hit");
hitsCollection = new ArrayList<Hit>();
for (Element hitElement : hitList) {
BlastHitBuilder blastHitBuilder = new BlastHitBuilder();
blastHitBuilder
.setHitNum(
new Integer(
XMLHelper.selectSingleElement(hitElement, "Hit_num").getTextContent()))
.setHitId(XMLHelper.selectSingleElement(hitElement, "Hit_id").getTextContent())
.setHitDef(XMLHelper.selectSingleElement(hitElement, "Hit_def").getTextContent())
.setHitAccession(
XMLHelper.selectSingleElement(hitElement, "Hit_accession").getTextContent())
.setHitLen(
new Integer(
XMLHelper.selectSingleElement(hitElement, "Hit_len").getTextContent()));
if (databaseReferences != null)
blastHitBuilder.setHitSequence(
databaseReferencesMap.get(
XMLHelper.selectSingleElement(hitElement, "Hit_id").getTextContent()));
Element hithspsElement = XMLHelper.selectSingleElement(hitElement, "Hit_hsps");
ArrayList<Element> hspList = XMLHelper.selectElements(hithspsElement, "Hsp");
hspsCollection = new ArrayList<Hsp>();
for (Element hspElement : hspList) {
Double evalue =
new Double(
XMLHelper.selectSingleElement(hspElement, "Hsp_evalue").getTextContent());
// add the new hsp only if it pass the specified threshold. It can save lot of memory
// and some parsing time
if (evalue <= maxEScore) {
BlastHspBuilder blastHspBuilder = new BlastHspBuilder();
blastHspBuilder
.setHspNum(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_num").getTextContent()))
.setHspBitScore(
new Double(
XMLHelper.selectSingleElement(hspElement, "Hsp_bit-score")
.getTextContent()))
.setHspScore(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_score").getTextContent()))
.setHspEvalue(evalue)
.setHspQueryFrom(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_query-from")
.getTextContent()))
.setHspQueryTo(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_query-to")
.getTextContent()))
.setHspHitFrom(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_hit-from")
.getTextContent()))
.setHspHitTo(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_hit-to").getTextContent()))
.setHspQueryFrame(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_query-frame")
.getTextContent()))
.setHspHitFrame(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_hit-frame")
.getTextContent()))
.setHspIdentity(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_identity")
.getTextContent()))
.setHspPositive(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_positive")
.getTextContent()))
.setHspGaps(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_gaps").getTextContent()))
.setHspAlignLen(
new Integer(
XMLHelper.selectSingleElement(hspElement, "Hsp_align-len")
.getTextContent()))
.setHspQseq(
XMLHelper.selectSingleElement(hspElement, "Hsp_qseq").getTextContent())
.setHspHseq(
XMLHelper.selectSingleElement(hspElement, "Hsp_hseq").getTextContent())
.setHspIdentityString(
XMLHelper.selectSingleElement(hspElement, "Hsp_midline").getTextContent());
hspsCollection.add(blastHspBuilder.createBlastHsp());
}
}
// finally set the computed hsp collection and create Hit object
blastHitBuilder.setHsps(hspsCollection);
hitsCollection.add(blastHitBuilder.createBlastHit());
}
// finally set the computed Hit collection to the result
resultBuilder.setHits(hitsCollection);
resultsCollection.add(resultBuilder.createBlastResult());
}
} catch (XPathException e) {
throw new ParseException(e.getMessage(), 0);
}
logger.info("Parsing of " + targetFile + " finished.");
return resultsCollection;
}
public GeneIDXMLReader(String geneidXMLFile) throws Exception {
logger.info("Start read of {}", geneidXMLFile);
geneidDoc = XMLHelper.loadXML(geneidXMLFile);
logger.info("Read finished");
}