/**
* Sample and calculate the probability of hitting each type of marker (marker.class). Creates
* 'numReads' reads of size 'readLen' and count how many of them hit each marker type.
*/
CountByType randomSampling(int readLen, int numReads) {
CountByType countReads = new CountByType();
RandMarker randMarker = new RandMarker(snpEffectPredictor.getGenome());
for (int i = 0; i < numReads; i++) {
// Random read
Marker read = randMarker.rand(readLen);
// Where does it hit?
Markers regions = snpEffectPredictor.queryDeep(read);
HashSet<String> doneRegion = new HashSet<String>();
for (Marker m : regions) {
String mtype = markerTypes.getType(m);
String msubtype = markerTypes.getSubType(m);
if (!doneRegion.contains(mtype)) {
countReads.inc(mtype); // Count reads
doneRegion.add(mtype); // Do not count twice
}
if ((msubtype != null) && !doneRegion.contains(msubtype)) {
countReads.inc(msubtype); // Count reads
doneRegion.add(msubtype); // Do not count twice
}
}
}
return countReads;
}
/** Count bases covered for each marker type */
public void countBases() {
// ---
// Add all markers
// ---
Markers markers = new Markers();
markers.add(snpEffectPredictor.getMarkers());
for (Gene gene : snpEffectPredictor.getGenome().getGenes()) {
markers.add(gene);
markers.add(gene.markers());
}
for (Chromosome chr : snpEffectPredictor.getGenome()) markers.add(chr);
// ---
// Calculate raw counts
// ---
for (Marker m : markers) {
String mtype = markerTypes.getType(m);
String msubtype = markerTypes.getSubType(m);
rawCountMarkers.inc(mtype);
rawCountBases.inc(mtype, m.size());
// Count sub-types (if any)
if (msubtype != null) {
rawCountMarkers.inc(msubtype);
rawCountBases.inc(msubtype, m.size());
}
}
// ---
// Count number of bases for each marker type (overlap and join)
// ---
for (String mtype : rawCountMarkers.keysSorted()) {
if (mtype.equals(Chromosome.class.getSimpleName()))
continue; // We calculate chromosomes later (it's faster)
if (verbose) System.err.print(mtype + ":");
if (countMarkers.get(mtype) == 0) {
for (Chromosome chr : snpEffectPredictor.getGenome()) countBases(mtype, chr, markers);
}
if (verbose) System.err.println("");
}
// Show chromosomes length
String mtype = Chromosome.class.getSimpleName();
for (Chromosome chr : snpEffectPredictor.getGenome()) {
countBases.inc(mtype, chr.size());
countMarkers.inc(mtype);
}
}
/**
* Count number of bases, for a given chromosome and marker type
*
* @param mtype
* @param chr
* @param markers
* @return
*/
void countBases(String mtype, Chromosome chr, Markers markers) {
String chrName = chr.getChromosomeName();
if (verbose) System.err.print(" " + chrName);
// Initialize
byte busy[] = new byte[chr.size()];
for (int i = 0; i < busy.length; i++) busy[i] = 0;
for (Marker m : markers) {
// Same marker type & same chromo? Count bases
if (m.getChromosomeName().equals(chrName) && markerTypes.isType(m, mtype)) {
for (int i = m.getStart(); i <= m.getEnd(); i++) busy[i] = 1;
}
}
int latest = 0;
for (int i = 0; i < busy.length; i++) {
// Transition? Count another marker
if ((i > 0) && (busy[i] != 0) && (busy[i - 1] == 0)) {
if ((i - latest) <= readLength)
countBases.inc(mtype, i - latest); // Intervals are less than one read away? Unify them
else countMarkers.inc(mtype);
}
// Base busy? Count another base
if (busy[i] != 0) {
countBases.inc(mtype);
latest = i;
}
}
}
/**
* Save model to file
*
* @param fileName
*/
public void save(String fileName) {
StringBuilder sb = new StringBuilder();
sb.append("marker_type\tsize\tcount\tbinomial_p\n");
probabilities();
for (String mtype : markerTypes.markerTypesClass())
sb.append(mtype + "\t" + countBases.get(mtype) + "\t" + countMarkers.get(mtype) + "\n");
Gpr.toFile(fileName, sb.toString());
}