/** Calculate probabilities */
void probabilities() {
// Already done, nothing to do
if (prob != null) return;
// Get total length and count for chromosomes (chromosome size is total genome length)
String chrType = Chromosome.class.getSimpleName();
long chrSize = countBases.get(chrType);
long chrCount = countMarkers.get(chrType);
if (chrCount <= 0) return; // Zero length genome? Forgot to count bases?
// Correct readLength
int readLength = this.readLength;
if (readLength < 1) readLength = 1;
// Probabilities for each marker
prob = new CountByType();
for (String mtype : countMarkers.keysSorted()) {
long size = countBases.get(mtype);
long count = countMarkers.get(mtype);
// Calculate and cap probability value
double p =
((double) (size + (readLength - 1) * count))
/ ((double) (chrSize - (readLength - 1) * chrCount));
p = Math.min(1.0, p);
p = Math.max(0.0, p);
prob.setScore(mtype, p);
}
}
/**
* 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());
}
/** 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);
}
}
/**
* 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. Iterate
* 'iterations' times to obtain a distribution.
*/
public void randomSampling(int iterations, int readLen, int numReads) {
System.out.print("Iteration");
for (String type : rawCountMarkers.keysSorted()) System.out.print("\t" + type);
System.out.println("");
for (int it = 0; it < iterations; it++) {
CountByType count = randomSampling(readLen, numReads);
System.out.print(it);
for (String type : rawCountMarkers.keysSorted()) System.out.print("\t" + count.get(type));
System.out.println("");
}
}
@Override
public String toString() {
StringBuilder sb = new StringBuilder();
sb.append("marker_type\tsize\tcount\traw_size\traw_count\tbinomial_p\n");
probabilities();
for (String mtype : countMarkers.keysSorted())
sb.append(
mtype
+ "\t"
+ countBases.get(mtype)
+ "\t"
+ countMarkers.get(mtype)
+ "\t"
+ rawCountBases.get(mtype)
+ "\t"
+ rawCountMarkers.get(mtype)
+ "\t"
+ prob.getScore(mtype)
+ "\n");
return sb.toString();
}