Modified build rules for sam/libbam.a to enable parallel build

[rsem.git] / PairedEndQModel.h

diff --git a/PairedEndQModel.h b/PairedEndQModel.h
index e328779b6675a3579aa221c32f490b99648053e4..2c638abed103bcf4f7167b70b60221aa9eafed88 100644 (file)
--- a/PairedEndQModel.h
+++ b/PairedEndQModel.h
@@ -61,12 +61,12 @@ public:
                mw = NULL;
 
                if (isMaster) {
-                       ori = new Orientation(params.probF);
                        if (!estRSPD) rspd = new RSPD(estRSPD);
                        qd = new QualDist();
                        mld = new LenDist(params.mate_minL, params.mate_maxL);
                }
 
+               ori = new Orientation(params.probF);
                gld = new LenDist(params.minL, params.maxL);
                if (estRSPD) rspd = new RSPD(estRSPD, params.B);
                qpro = new QProfile();
@@ -199,8 +199,6 @@ public:
        }
 
        int getModelType() const { return model_type; }
-
-       bool simulate(int, PairedEndReadQ&, SingleReadQ&, int&);
  
 private:
        static const int model_type = 3;
@@ -238,22 +236,27 @@ void PairedEndQModel::estimateFromReads(const char* readFN) {
     for (int i = 0; i < 3; i++)
        if (N[i] > 0) {
                genReadFileNames(readFN, i, read_type, s, readFs);
-               ReadReader<PairedEndReadQ> reader(s, readFs);
+               ReadReader<PairedEndReadQ> reader(s, readFs, refs->hasPolyA(), seedLen); // allow calculation of calc_lq() function
 
                int cnt = 0;
                while (reader.next(read)) {
                        SingleReadQ mate1 = read.getMate1();
                        SingleReadQ mate2 = read.getMate2();
 
-                       mld->update(mate1.getReadLength(), 1.0);
-                       mld->update(mate2.getReadLength(), 1.0);
-
-                       qd->update(mate1.getQScore());
-                       qd->update(mate2.getQScore());
-
-                       if (i == 0) {
-                               nqpro->updateC(mate1.getReadSeq(), mate1.getQScore());
-                               nqpro->updateC(mate2.getReadSeq(), mate2.getQScore());
+                       if (!read.isLowQuality()) {
+                               mld->update(mate1.getReadLength(), 1.0);
+                               mld->update(mate2.getReadLength(), 1.0);
+
+                               qd->update(mate1.getQScore());
+                               qd->update(mate2.getQScore());
+                         
+                               if (i == 0) {
+                                       nqpro->updateC(mate1.getReadSeq(), mate1.getQScore());
+                                       nqpro->updateC(mate2.getReadSeq(), mate2.getQScore());
+                               }
+                       }
+                       else if (verbose && (mate1.getReadLength() < seedLen || mate2.getReadLength() < seedLen)) {
+                               printf("Warning: Read %s is ignored due to at least one of the mates' length < seed length %d!\n", read.getName().c_str(), seedLen);
                        }
 
                        ++cnt;
@@ -300,7 +303,7 @@ void PairedEndQModel::read(const char* inpF) {
        FILE *fi = fopen(inpF, "r");
        if (fi == NULL) { fprintf(stderr, "Cannot open %s! It may not exist.\n", inpF); exit(-1); }
 
-       fscanf(fi, "%d", &val);
+       assert(fscanf(fi, "%d", &val) == 1);
        assert(val == model_type);
 
        ori->read(fi);
@@ -311,15 +314,19 @@ void PairedEndQModel::read(const char* inpF) {
        qpro->read(fi);
        nqpro->read(fi);
 
-       if (fscanf(fi, "%d", &M) == 1) {
-         mw = new double[M + 1];
-         for (int i = 0; i <= M; i++) fscanf(fi, "%lf", &mw[i]);
+       if (fscanf(fi, "%d", &val) == 1) {
+               if (M == 0) M = val;
+               if (M == val) {
+                       mw = new double[M + 1];
+                       for (int i = 0; i <= M; i++) assert(fscanf(fi, "%lf", &mw[i]) == 1);
+               }
        }
 
+
        fclose(fi);
 }
 
-//Only master node can call
+//Only master node can call. Only be called at EM.cpp
 void PairedEndQModel::write(const char* outF) {
        FILE *fo = fopen(outF, "w");
 
@@ -335,7 +342,7 @@ void PairedEndQModel::write(const char* outF) {
        nqpro->write(fo);
 
        if (mw != NULL) {
-         fprintf(fo, "%d\n", M);
+         fprintf(fo, "\n%d\n", M);
          for (int i = 0; i < M; i++) {
            fprintf(fo, "%.15g ", mw[i]);
          }
@@ -401,9 +408,8 @@ bool PairedEndQModel::simulate(int rid, PairedEndReadQ& read, int& sid) {
                readseq2 = qpro->simulate(sampler, mateL2, m2pos, m2dir, qual2, ref);
        }
 
-       std::ostringstream stdout;
-       stdout<<rid<<"_"<<dir<<"_"<<sid<<"_"<<pos<<"_"<<insertL;
-       name = stdout.str();
+       strout<<rid<<"_"<<dir<<"_"<<sid<<"_"<<pos<<"_"<<insertL;
+       name = strout.str();
 
        read = PairedEndReadQ(SingleReadQ(name + "/1", readseq1, qual1), SingleReadQ(name + "/2", readseq2, qual2));
 
@@ -421,39 +427,39 @@ void PairedEndQModel::finishSimulation() {
 
 
 void PairedEndQModel::calcMW() {
-  assert(seedLen >= OLEN && mld->getMinL() >= seedLen);
-
-  memset(mw, 0, sizeof(double) * (M + 1));
-  mw[0] = 1.0;
-
-  for (int i = 1; i <= M; i++) { 
-    RefSeq& ref = refs->getRef(i);
-    int totLen = ref.getTotLen();
-    int fullLen = ref.getFullLen();
-    int end = std::min(fullLen, totLen - gld->getMinL() + 1);
-    double value = 0.0;
-    int minL, maxL;
-    int effL, pfpos;
-
-    //seedPos is fpos here
-    for (int seedPos = 0; seedPos < end; seedPos++) 
-      if (ref.getMask(seedPos)) {
-       minL = gld->getMinL();
-       maxL = std::min(gld->getMaxL(), totLen - seedPos);
-       pfpos = seedPos;
-       for (int fragLen = minL; fragLen <= maxL; fragLen++) {
-         effL = std::min(fullLen, totLen - fragLen + 1); 
-         value += gld->getAdjustedProb(fragLen, totLen) * rspd->getAdjustedProb(pfpos, effL, fullLen); 
-       }
-      }
+       assert(mld->getMinL() >= seedLen);
+
+       memset(mw, 0, sizeof(double) * (M + 1));
+       mw[0] = 1.0;
+
+       for (int i = 1; i <= M; i++) {
+               RefSeq& ref = refs->getRef(i);
+               int totLen = ref.getTotLen();
+               int fullLen = ref.getFullLen();
+               int end = std::min(fullLen, totLen - gld->getMinL() + 1);
+               double value = 0.0;
+               int minL, maxL;
+               int effL, pfpos;
+
+               //seedPos is fpos here
+               for (int seedPos = 0; seedPos < end; seedPos++)
+                       if (ref.getMask(seedPos)) {
+                               minL = gld->getMinL();
+                               maxL = std::min(gld->getMaxL(), totLen - seedPos);
+                               pfpos = seedPos;
+                               for (int fragLen = minL; fragLen <= maxL; fragLen++) {
+                                       effL = std::min(fullLen, totLen - fragLen + 1);
+                                       value += gld->getAdjustedProb(fragLen, totLen) * rspd->getAdjustedProb(pfpos, effL, fullLen);
+                               }
+                       }
     
-    mw[i] = 1.0 - value;
+               mw[i] = 1.0 - value;
 
-    if (mw[i] < 1e-8) { 
-      //      fprintf(stderr, "Warning: %dth reference sequence is masked for almost all positions!\n", i);
-      mw[i] = 0.0;
-    }
-  }
+               if (mw[i] < 1e-8) {
+                       //      fprintf(stderr, "Warning: %dth reference sequence is masked for almost all positions!\n", i);
+                       mw[i] = 0.0;
+               }
+       }
 }
 
 #endif /* PAIREDENDQMODEL_H_ */