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;
}
public NamedRegion createNamedRegion(BlastTabEntry e) {
int start = Math.min(e.getSubjectStart(), e.getSubjectEnd());
int end = Math.max(e.getSubjectStart(), e.getSubjectEnd());
String chrom = getChrom(e.getSubject());
return new NamedRegion(genome, chrom, start, end, e.getQuery());
}
public static void main(String args[]) throws Exception {
Genome genome = Args.parseGenome(args).cdr();
ChipChipDataset dataset = new ChipChipDataset(genome);
List<Region> regions = Args.parseRegionsOrDefault(args);
double means[] = new double[CGHCallExpander.numStates],
stddevs[] = new double[CGHCallExpander.numStates];
String[] pieces =
Args.parseString(
args,
"low",
String.format(
"%f;%f",
CGHCallExpander.defaultParams[0][0],
Math.sqrt(CGHCallExpander.defaultParams[0][1])))
.split(";");
means[CGHCallExpander.LOW] = Double.parseDouble(pieces[0]);
stddevs[CGHCallExpander.LOW] = Math.pow(Double.parseDouble(pieces[1]), 2);
pieces =
Args.parseString(
args,
"middle",
String.format(
"%f;%f",
CGHCallExpander.defaultParams[0][0],
Math.sqrt(CGHCallExpander.defaultParams[0][1])))
.split(";");
means[CGHCallExpander.ONE] = Double.parseDouble(pieces[0]);
stddevs[CGHCallExpander.ONE] = Math.pow(Double.parseDouble(pieces[1]), 2);
pieces =
Args.parseString(
args,
"high",
String.format(
"%f;%f",
CGHCallExpander.defaultParams[0][0],
Math.sqrt(CGHCallExpander.defaultParams[0][1])))
.split(";");
means[CGHCallExpander.HIGH] = Double.parseDouble(pieces[0]);
stddevs[CGHCallExpander.HIGH] = Math.pow(Double.parseDouble(pieces[1]), 2);
for (ExptNameVersion env : Args.parseENV(args)) {
ChipChipData ccd = dataset.getData(env);
CGHCallExpander cgh = new CGHCallExpander(ccd);
cgh.setStateParameters(
CGHCallExpander.LOW, means[CGHCallExpander.LOW], stddevs[CGHCallExpander.LOW]);
cgh.setStateParameters(
CGHCallExpander.ONE, means[CGHCallExpander.ONE], stddevs[CGHCallExpander.ONE]);
cgh.setStateParameters(
CGHCallExpander.HIGH, means[CGHCallExpander.HIGH], stddevs[CGHCallExpander.HIGH]);
cgh.resetHMM();
for (Region r : regions) {
List<ScoredRegion> calls = cgh.executeList(r);
for (ScoredRegion call : calls) {
System.out.println(call.getScore() + "\t" + call.regionString());
}
}
}
}