/** {@inheritDoc} */
public void process(Reader reader) throws ObjectStoreException, IOException {
initialiseMapsForFile();
BufferedReader br = new BufferedReader(reader);
String line = null;
// loop through entire file
while ((line = br.readLine()) != null) {
if (line.startsWith("!")) {
continue;
}
String[] array = line.split("\t", -1); // keep trailing empty Strings
if (array.length < 13) {
throw new IllegalArgumentException(
"Not enough elements (should be > 13 not " + array.length + ") in line: " + line);
}
String taxonId = parseTaxonId(array[12]);
Config config = configs.get(taxonId);
if (config == null) {
throw new NullPointerException(
"No entry for organism with taxonId = '" + taxonId + "' found in config file.");
}
int readColumn = config.readColumn();
String productId = array[readColumn];
String goId = array[4];
String qualifier = array[3];
String strEvidence = array[6];
String withText = array[7];
if (StringUtils.isNotEmpty(strEvidence)) {
storeEvidenceCode(strEvidence);
} else {
throw new IllegalArgumentException(
"Evidence is a required column but not "
+ "found for doterm "
+ goId
+ " and productId "
+ productId);
}
// type of gene product, we're not interested in at the moment
// String type = array[11];
String type = configs.get(taxonId).annotationType;
// Wormbase has some proteins with UniProt accessions and some with WB:WP ids,
// hack here to get just the UniProt ones.
// if (("protein".equalsIgnoreCase(type) && !array[0].startsWith("UniProt"))
// || (!"protein".equalsIgnoreCase(type) && array[0].startsWith("UniProt"))) {
// continue;
// }
// create unique key for go annotation
GoTermToGene key = new GoTermToGene(productId, goId, qualifier);
String dataSourceCode = array[14];
Item organism = newOrganism(taxonId);
String productIdentifier = newProduct(productId, type, organism, dataSourceCode, true, null);
// null if resolver could not resolve an identifier
if (productIdentifier != null) {
// null if no pub found
String pubRefId = newPublication(array[5]);
// get evidence codes for this goterm|gene pair
Set<Evidence> allEvidenceForAnnotation = goTermGeneToEvidence.get(key);
// new evidence
if (allEvidenceForAnnotation == null) {
String goTermIdentifier = newGoTerm(goId, dataSourceCode);
Evidence evidence = new Evidence(strEvidence, pubRefId);
allEvidenceForAnnotation = new LinkedHashSet<Evidence>();
allEvidenceForAnnotation.add(evidence);
goTermGeneToEvidence.put(key, allEvidenceForAnnotation);
Integer storedAnnotationId =
createGoAnnotation(
productIdentifier,
type,
goTermIdentifier,
organism,
qualifier,
withText,
dataSourceCode);
evidence.setStoredAnnotationId(storedAnnotationId);
} else {
boolean seenEvidenceCode = false;
Integer storedAnnotationId = null;
for (Evidence evidence : allEvidenceForAnnotation) {
String evidenceCode = evidence.getEvidenceCode();
// already have evidence code, just add pub
if (evidenceCode.equals(strEvidence)) {
evidence.addPublicationRefId(pubRefId);
seenEvidenceCode = true;
}
storedAnnotationId = evidence.storedAnnotationId;
}
if (!seenEvidenceCode) {
Evidence evidence = new Evidence(strEvidence, pubRefId);
evidence.storedAnnotationId = storedAnnotationId;
allEvidenceForAnnotation.add(evidence);
}
}
}
}
storeProductCollections();
storeEvidence();
}
