private Pair<Region, Region> findProbePair(
String primerName, Map<String, Set<Region>> primerHits) {
String leftName = String.format("%s_left", primerName);
String rightName = String.format("%s_right", primerName);
if (!primerHits.containsKey(leftName) || !primerHits.containsKey(rightName)) {
return null;
}
if (primerHits.get(leftName).size() != 1 || primerHits.get(rightName).size() != 1) {
return null;
}
Iterator<Region> leftitr = primerHits.get(leftName).iterator();
Iterator<Region> rightitr = primerHits.get(rightName).iterator();
Region left = leftitr.next();
Region right = rightitr.next();
if (!left.getChrom().equals(right.getChrom())) {
return null;
}
return new Pair<Region, Region>(left, right);
}
public Collection<NamedRegion> parseProbes() throws IOException {
double ethreshold = 1.0e-3;
Parser<BlastTabEntry> parser =
new Parser<BlastTabEntry>(blasttab, new BlastTabEntry.ParsingMapper());
Iterator<BlastTabEntry> itr =
new FilterIterator<BlastTabEntry, BlastTabEntry>(
new BlastTabEntry.ExpectedScoreFilter(ethreshold), parser);
Map<String, Set<Region>> primerHits = new TreeMap<String, Set<Region>>();
Set<String> primerNames = new TreeSet<String>();
while (itr.hasNext()) {
BlastTabEntry entry = itr.next();
NamedRegion r = createNamedRegion(entry);
if (!primerHits.containsKey(entry.getQuery())) {
primerHits.put(entry.getQuery(), new HashSet<Region>());
}
primerHits.get(entry.getQuery()).add(r);
primerNames.add(removePrimerLR(entry.getQuery()));
}
LinkedList<NamedRegion> probes = new LinkedList<NamedRegion>();
for (String primerName : primerNames) {
Pair<Region, Region> hits = findProbePair(primerName, primerHits);
if (hits != null) {
Region left = hits.getFirst(), right = hits.getLast();
int start = Math.min(left.getStart(), right.getStart());
int end = Math.max(left.getEnd(), right.getEnd());
NamedRegion probe = new NamedRegion(genome, left.getChrom(), start, end, primerName);
probes.addLast(probe);
}
}
return probes;
}