rsem v1.1.14, add --sampling-for-bam option, modify rsem-bam2wig to handle BAM files...

[rsem.git] / PairedEndQModel.h

diff --git a/PairedEndQModel.h b/PairedEndQModel.h
index bac7135721244a704475e73a6a010a989375d2b7..cbc5475588d19996154d892cc99af336b75ddbde 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;
@@ -245,16 +243,21 @@ void PairedEndQModel::estimateFromReads(const char* readFN) {
                        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() < OLEN || mate2.getReadLength() < OLEN)) { 
+                         printf("Warning: Read %s is ignored due to at least one of the mates' length < %d!\n", read.getName().c_str(), OLEN);
+                       }
 
                        ++cnt;
                        if (verbose && cnt % 1000000 == 0) { printf("%d READS PROCESSED\n", 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");
 
@@ -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));