/**
* Return a qualifier copy of the qualifier provided that does not contain any of the values in
* the StringVector.
*
* @param newQualifier
* @param oldValues
* @return
* @throws InvalidRelationException
*/
protected static Qualifier getQualifierWithoutDuplicateValues(
final Qualifier qualifier, final StringVector values) throws InvalidRelationException {
final Qualifier newQualifier;
if (values == null || values.size() < 1) newQualifier = qualifier;
else {
StringVector newValues = qualifier.getValues();
StringVector valuesToAdd = new StringVector();
for (int k = 0; k < newValues.size(); k++) {
if (!values.contains(newValues.get(k))) {
if (qualifier.getName().equals("history")) {
if (!uk.ac.sanger.artemis.components.genebuilder.cv.HistoryBox.contains(
values, (String) newValues.get(k))) valuesToAdd.add(newValues.get(k));
} else valuesToAdd.add(newValues.get(k));
}
}
if (valuesToAdd.size() == 0) return null;
newQualifier = new Qualifier(qualifier.getName(), valuesToAdd);
}
return newQualifier;
}
/**
* Optionally transfer GO fields with evidence code ISO and link back to the original source in
* the WITH/FROM column.
*
* @param setEvidenceAndWithFrom
* @param feature
* @param qName
* @param values
* @return
*/
private static StringVector getTransferValues(
final boolean setEvidenceAndWithFrom,
final Feature feature,
final String qName,
final StringVector values) {
if (!setEvidenceAndWithFrom) return values;
if (qName.equals("GO") || qName.equals("product")) {
final StringVector tvalues = new StringVector();
final String gene = getGeneName(feature);
for (int i = 0; i < values.size(); i++) {
String val =
changeField("evidence=", "Inferred from Sequence Orthology", null, values.get(i));
if (gene != null) val = changeField("with=", "GeneDB:" + gene, "|", val);
tvalues.add(val);
}
return tvalues;
}
return values;
}
/**
* Merge qualifiers
*
* @param qualifiers
* @param newQualifiers
*/
private void combineQualifiers(
final QualifierVector qualifiers, final QualifierVector newQualifiers, final boolean isGene) {
for (int i = 0; i < newQualifiers.size(); i++) {
Qualifier newQualifier = (Qualifier) newQualifiers.get(i);
if (newQualifier.getName().equals("ID") && !isGene) {
continue;
}
// convert GO evidence to codes (e.g. ND=No biological Data available)
if (newQualifier.getName().equals("GO")) {
final StringVector newValues = newQualifier.getValues();
final StringVector tmpNewValues = new StringVector();
for (int j = 0; j < newValues.size(); j++) {
String val = GoBox.getEvidenceCodeGoTextFromText((String) newValues.get(j));
tmpNewValues.add(val);
}
newQualifier = new Qualifier("GO", tmpNewValues);
}
if (newQualifier.getName().equals("product")) {
final StringVector newValues = newQualifier.getValues();
final StringVector tmpNewValues = new StringVector();
for (int j = 0; j < newValues.size(); j++) {
String val = (String) newValues.get(j);
int ind = 0;
if ((ind = val.indexOf(";db_xref=")) > -1) val = val.substring(0, ind);
if ((ind = val.indexOf(";evidence=")) > -1) val = val.substring(0, ind);
if (val.startsWith("term=")) val = val.substring(5, val.length());
if (val.endsWith(";")) val = val.substring(0, val.length() - 1);
tmpNewValues.add(val);
}
newQualifier = new Qualifier("product", tmpNewValues);
}
if (newQualifier.getName().equals("orthologous_to")
|| newQualifier.getName().equals("paralogous_to")) {
final StringVector newValues = newQualifier.getValues();
final StringVector tmpNewValues = new StringVector();
for (int j = 0; j < newValues.size(); j++) {
if (!newValues.get(j).equals("")) tmpNewValues.add(newValues.get(j));
}
if (tmpNewValues.size() == 0) continue;
Pattern p = Pattern.compile("\\w+:link=\\w+");
for (int j = 0; j < tmpNewValues.size(); j++) {
String valueStr = (String) tmpNewValues.get(j);
String newValueStr;
int indexEnd = valueStr.indexOf(';');
String endStr = "";
if (indexEnd > -1) endStr = valueStr.substring(indexEnd);
Matcher m = p.matcher(valueStr);
while (m.find()) {
int index = valueStr.indexOf("link=", m.start());
newValueStr =
valueStr.substring(m.start(), index)
+ valueStr.substring(index + 5, m.end())
+ endStr;
if (newQualifier.getName().equals("orthologous_to"))
newQualifier = new Qualifier("orthologous_to", newValueStr);
else newQualifier = new Qualifier("paralogous_to", newValueStr);
qualifiers.addElement(newQualifier);
}
}
continue;
}
addNewQualifier(qualifiers, newQualifier);
}
}
/**
* Map GFF features to EMBL/Genbank
*
* @param feature
* @return
*/
private Object mapGffToNativeFeature(final Feature feature) {
if (DATABASE_MAP_KEYS == null) initDatabaseMappings();
Key key = feature.getKey();
QualifierVector qualifiers = feature.getQualifiers().copy();
// ignore if obsolete
if (IGNORE_OBSOLETE_FEATURES) {
Qualifier isObsoleteQualifier = qualifiers.getQualifierByName("isObsolete");
if (isObsoleteQualifier != null) {
String value = (String) isObsoleteQualifier.getValues().get(0);
if (Boolean.parseBoolean(value)) return null;
}
}
key = map(key, qualifiers);
if (getEntryInformation().isValidQualifier((String) DATABASE_QUALIFIERS_TO_REMOVE[0])) {
try {
if (this instanceof EmblDocumentEntry)
return new EmblStreamFeature(key, feature.getLocation(), qualifiers);
else return new GenbankStreamFeature(key, feature.getLocation(), qualifiers);
} catch (InvalidRelationException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
Location location = joinUtrs(feature, key, qualifiers);
if (location == null) return null;
// flatten gene model - combining qualifiers
if (key.getKeyString().equals(DatabaseDocument.EXONMODEL)) {
ChadoCanonicalGene chadoGene = ((GFFStreamFeature) feature).getChadoGene();
final String name = GeneUtils.getUniqueName(feature);
final String transcriptName = chadoGene.getTranscriptFromName(name);
StringVector sv = new StringVector();
sv.add(transcriptName);
final Feature transcript = chadoGene.containsTranscript(sv);
if (transcript != null && GeneUtils.isNonCodingTranscripts(transcript.getKey())) return null;
qualifiers.removeQualifierByName("ID");
int ntranscripts = 0;
// add transcript & protein qualifiers to CDS
try {
final Feature protein = chadoGene.getProteinOfTranscript(transcriptName);
if (protein != null) combineQualifiers(qualifiers, protein.getQualifiers().copy(), false);
if (transcript != null)
ntranscripts = handleTranscripts(qualifiers, transcript, ntranscripts, chadoGene);
} catch (NullPointerException npe) {
}
// add gene qualifiers to CDS
QualifierVector geneQualifiers = chadoGene.getGene().getQualifiers().copy();
// multiple transcripts
if (ntranscripts > 1 && geneQualifiers.getQualifierByName("ID") != null) {
Qualifier newIDQualifier =
new Qualifier(
"shared_id", (String) geneQualifiers.getQualifierByName("ID").getValues().get(0));
addNewQualifier(qualifiers, newIDQualifier);
geneQualifiers.removeQualifierByName("ID");
}
combineQualifiers(qualifiers, geneQualifiers, true);
} else if (GeneUtils.isNonCodingTranscripts(key)) {
// use gene id for non-coding transcripts
ChadoCanonicalGene chadoGene = ((GFFStreamFeature) feature).getChadoGene();
if (chadoGene != null) {
qualifiers.removeQualifierByName("ID");
QualifierVector geneQualifiers = chadoGene.getGene().getQualifiers().copy();
combineQualifiers(qualifiers, geneQualifiers, true);
}
}
try {
for (int i = 0; i < DATABASE_QUALIFIERS_TO_MAP.length; i++) {
if (!getEntryInformation().isValidQualifier(DATABASE_QUALIFIERS_TO_MAP[i][0])) {
changeQualifierName(
qualifiers, DATABASE_QUALIFIERS_TO_MAP[i][0], DATABASE_QUALIFIERS_TO_MAP[i][1]);
}
}
if (qualifiers.getQualifierByName("stop_codon_redefined_as_selenocysteine") != null) {
handleSelenocysteine(qualifiers, feature);
}
for (int i = 0; i < DATABASE_QUALIFIERS_TO_REMOVE.length; i++) {
if (!getEntryInformation().isValidQualifier((String) DATABASE_QUALIFIERS_TO_REMOVE[i]))
qualifiers.removeQualifierByName((String) DATABASE_QUALIFIERS_TO_REMOVE[i]);
}
if (key.getKeyString().equals("polypeptide")) return null;
else if (key.getKeyString().equals("gene")) return null;
else if (key.getKeyString().equals("centromere")) return null;
else if (key.getKeyString().equals("transcript") || key.getKeyString().equals("mRNA"))
return null;
if (this instanceof EmblDocumentEntry)
return new EmblStreamFeature(key, location, qualifiers);
else return new GenbankStreamFeature(key, location, qualifiers);
} catch (InvalidRelationException e) {
e.printStackTrace();
if (feature instanceof DatabaseStreamFeature) return new EmblStreamFeature();
else return new GenbankStreamFeature();
}
}