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;
}
@Override
public void extractFeatures(List regions) throws FeatureExtractorException {
for (Region r : (List<Region>) regions) {
if (r.hasAttribute(f1) && r.hasAttribute(f2)) {
Contour c1, c2;
try {
c1 = ContourUtils.getSubContour((Contour) r.getAttribute(f1), r.getStart(), r.getEnd());
c2 = ContourUtils.getSubContour((Contour) r.getAttribute(f2), r.getStart(), r.getEnd());
} catch (AuToBIException e) {
throw new FeatureExtractorException(e.getMessage());
}
r.setAttribute("rmse[" + f1 + "," + f2 + "]", contourRMSE(c1, c2));
r.setAttribute("meanError[" + f1 + "," + f2 + "]", contourError(c1, c2));
}
}
}
@Override
public void write(JsonWriter out, Region value) throws IOException {
out.beginObject()
.name("start")
.value(value.getStart() + 1)
.name("count")
.value(value.getCount())
.name("path")
.value(value.getSourcePath())
.name("commit")
.value(ObjectId.toString(value.getSourceCommit()))
.name("author")
.beginObject()
// TODO(dborowitz): Use an adapter from CommitJsonData instead.
.name("name")
.value(value.getSourceAuthor().getName())
.name("email")
.value(value.getSourceAuthor().getEmailAddress())
.name("time")
.value(df.format(value.getSourceAuthor()))
.endObject()
.endObject();
}
