/**
* parse parameters of all entries
*
* @param in
*/
private svm_problem parsePara(ArrayList<TrainingEntry> in, String taxo) {
int num = 0;
for (TrainingEntry t : in) num += t.NumOfMi();
svm_node[][] data = new svm_node[num][];
int i = 0;
for (TrainingEntry mi : in) {
pri = mi.getPriRNA();
pri.setPriPara();
for (int j = 0; j < mi.NumOfMi(); j++) {
pri.process(mi.getStart(j), mi.getSize(j), mi.getStrand(j));
ps = new ParaSummary(pri, pri.getPreRNA(), pri.getMiRNA());
data[i++] = svm.predictMatrix(mi.getLabel(j), ps.paraMatrix(taxo));
}
}
return svm.trainMatrix(data);
}
public ArrayList<TrainingEntry> positiveData(ArrayList<MiRBaseEntry> milist, int subnum) {
System.out.print("Reads miRBase data ");
ArrayList<TrainingEntry> pos = new ArrayList<TrainingEntry>();
int num = milist.size();
ArrayList<Integer> rmis = randomization(num, num); // randomizate the miRBase entries
MfeFold doFold;
int evNum = 0; // number of experimental verified miRNAs
int uevNum = 0; // number of unexperimental verified miRNAs
int positive = 0; // number of positive mirna
int n = 0;
for (Integer m : rmis) {
if (subnum > 0 && positive >= subnum) break; // get subnum mirans
MiRBaseEntry mi = milist.get(m);
TrainingEntry te = new TrainingEntry();
pri = new PriMiRNA(mi.getPriAccession(), mi.getPriSequence());
// fold pri-miRNA
doFold = new MfeFold(pri.getSeq());
doFold.cal();
pri.setStr(doFold.getStructure());
pri.setEnergy(doFold.getEnergy());
te.setPriRNA(pri); // store pri
int end5 = pri.getStr().lastIndexOf("(") + 1; // 5' end position of hairpin,count from 1
int start3 = pri.getStr().indexOf(")") + 1; // 3' start position of hairpin,count from 1
for (int i = 0; i < mi.getMatNum(); i++) {
// the miRNA should be experimental verified miRNA
if (mi.getEvidence(i).equals("experimental")) {
evNum += 1;
// the miRNA should not have two or more loops
if (end5 >= start3) continue;
int strand;
if (mi.getMatEnd(i) <= end5) strand = 5; // lie in 5'
else if (mi.getMatStart(i) >= start3) strand = 3; // lie in 3'
else continue; // the miRNA should not lie in loop area
te.addMiEntry(mi.getMatStart(i) - 1, mi.getMatEnd(i) - mi.getMatStart(i) + 1, strand, 1);
// for the fasta format mirna data
// te.setMiId(mi.getMatID(i));
// p_fas.put(te.getMiId(te.NumOfMi()-1), te.getMiSeq(te.NumOfMi()-1));
positive++;
} else uevNum += 1;
}
// add positive data
if (te.NumOfMi() > 0) pos.add(te);
n++;
System.out.print(n + ":" + mi.getPriID() + backspace(n + ":" + mi.getPriID()));
}
System.out.println();
// report
System.out.println("Loads " + (evNum + uevNum) + " Entries");
System.out.println("Experimental verified Entries: " + evNum);
System.out.println("Non-experimental verified Entries: " + uevNum);
System.out.println("Positive Entries: " + positive);
return pos;
}